Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
/**
|
2022-10-18 23:19:24 +08:00
|
|
|
* Copyright (c) Meta Platforms, Inc. and affiliates.
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
*
|
|
|
|
|
* This source code is licensed under the MIT license found in the
|
|
|
|
|
* LICENSE file in the root directory of this source tree.
|
|
|
|
|
*
|
|
|
|
|
* @flow
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
// UpdateQueue is a linked list of prioritized updates.
|
|
|
|
|
//
|
|
|
|
|
// Like fibers, update queues come in pairs: a current queue, which represents
|
2019-01-03 04:53:05 +08:00
|
|
|
// the visible state of the screen, and a work-in-progress queue, which can be
|
|
|
|
|
// mutated and processed asynchronously before it is committed — a form of
|
|
|
|
|
// double buffering. If a work-in-progress render is discarded before finishing,
|
|
|
|
|
// we create a new work-in-progress by cloning the current queue.
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
//
|
|
|
|
|
// Both queues share a persistent, singly-linked list structure. To schedule an
|
|
|
|
|
// update, we append it to the end of both queues. Each queue maintains a
|
|
|
|
|
// pointer to first update in the persistent list that hasn't been processed.
|
|
|
|
|
// The work-in-progress pointer always has a position equal to or greater than
|
|
|
|
|
// the current queue, since we always work on that one. The current queue's
|
|
|
|
|
// pointer is only updated during the commit phase, when we swap in the
|
|
|
|
|
// work-in-progress.
|
|
|
|
|
//
|
|
|
|
|
// For example:
|
|
|
|
|
//
|
|
|
|
|
// Current pointer: A - B - C - D - E - F
|
|
|
|
|
// Work-in-progress pointer: D - E - F
|
|
|
|
|
// ^
|
|
|
|
|
// The work-in-progress queue has
|
|
|
|
|
// processed more updates than current.
|
|
|
|
|
//
|
|
|
|
|
// The reason we append to both queues is because otherwise we might drop
|
|
|
|
|
// updates without ever processing them. For example, if we only add updates to
|
|
|
|
|
// the work-in-progress queue, some updates could be lost whenever a work-in
|
|
|
|
|
// -progress render restarts by cloning from current. Similarly, if we only add
|
|
|
|
|
// updates to the current queue, the updates will be lost whenever an already
|
|
|
|
|
// in-progress queue commits and swaps with the current queue. However, by
|
|
|
|
|
// adding to both queues, we guarantee that the update will be part of the next
|
|
|
|
|
// work-in-progress. (And because the work-in-progress queue becomes the
|
|
|
|
|
// current queue once it commits, there's no danger of applying the same
|
|
|
|
|
// update twice.)
|
|
|
|
|
//
|
|
|
|
|
// Prioritization
|
|
|
|
|
// --------------
|
|
|
|
|
//
|
|
|
|
|
// Updates are not sorted by priority, but by insertion; new updates are always
|
|
|
|
|
// appended to the end of the list.
|
|
|
|
|
//
|
|
|
|
|
// The priority is still important, though. When processing the update queue
|
|
|
|
|
// during the render phase, only the updates with sufficient priority are
|
|
|
|
|
// included in the result. If we skip an update because it has insufficient
|
|
|
|
|
// priority, it remains in the queue to be processed later, during a lower
|
|
|
|
|
// priority render. Crucially, all updates subsequent to a skipped update also
|
|
|
|
|
// remain in the queue *regardless of their priority*. That means high priority
|
|
|
|
|
// updates are sometimes processed twice, at two separate priorities. We also
|
|
|
|
|
// keep track of a base state, that represents the state before the first
|
|
|
|
|
// update in the queue is applied.
|
|
|
|
|
//
|
|
|
|
|
// For example:
|
|
|
|
|
//
|
|
|
|
|
// Given a base state of '', and the following queue of updates
|
|
|
|
|
//
|
|
|
|
|
// A1 - B2 - C1 - D2
|
|
|
|
|
//
|
|
|
|
|
// where the number indicates the priority, and the update is applied to the
|
|
|
|
|
// previous state by appending a letter, React will process these updates as
|
|
|
|
|
// two separate renders, one per distinct priority level:
|
|
|
|
|
//
|
|
|
|
|
// First render, at priority 1:
|
|
|
|
|
// Base state: ''
|
|
|
|
|
// Updates: [A1, C1]
|
|
|
|
|
// Result state: 'AC'
|
|
|
|
|
//
|
|
|
|
|
// Second render, at priority 2:
|
|
|
|
|
// Base state: 'A' <- The base state does not include C1,
|
|
|
|
|
// because B2 was skipped.
|
|
|
|
|
// Updates: [B2, C1, D2] <- C1 was rebased on top of B2
|
|
|
|
|
// Result state: 'ABCD'
|
|
|
|
|
//
|
|
|
|
|
// Because we process updates in insertion order, and rebase high priority
|
|
|
|
|
// updates when preceding updates are skipped, the final result is deterministic
|
|
|
|
|
// regardless of priority. Intermediate state may vary according to system
|
|
|
|
|
// resources, but the final state is always the same.
|
|
|
|
|
|
2021-01-28 03:55:27 +08:00
|
|
|
import type {Fiber, FiberRoot} from './ReactInternalTypes';
|
2022-12-02 12:19:13 +08:00
|
|
|
import type {Lanes, Lane} from './ReactFiberLane';
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
|
2020-12-05 00:54:09 +08:00
|
|
|
import {
|
|
|
|
|
NoLane,
|
|
|
|
|
NoLanes,
|
2022-06-30 11:31:42 +08:00
|
|
|
OffscreenLane,
|
2020-12-05 00:54:09 +08:00
|
|
|
isSubsetOfLanes,
|
|
|
|
|
mergeLanes,
|
2022-06-30 11:31:42 +08:00
|
|
|
removeLanes,
|
2021-01-28 03:55:27 +08:00
|
|
|
isTransitionLane,
|
|
|
|
|
intersectLanes,
|
|
|
|
|
markRootEntangled,
|
2022-12-02 12:19:13 +08:00
|
|
|
} from './ReactFiberLane';
|
2019-01-25 03:31:20 +08:00
|
|
|
import {
|
|
|
|
|
enterDisallowedContextReadInDEV,
|
|
|
|
|
exitDisallowedContextReadInDEV,
|
2022-12-02 12:19:13 +08:00
|
|
|
} from './ReactFiberNewContext';
|
2022-07-08 23:51:40 +08:00
|
|
|
import {
|
|
|
|
|
Callback,
|
|
|
|
|
Visibility,
|
|
|
|
|
ShouldCapture,
|
|
|
|
|
DidCapture,
|
|
|
|
|
} from './ReactFiberFlags';
|
2022-12-03 01:14:12 +08:00
|
|
|
import getComponentNameFromFiber from './getComponentNameFromFiber';
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
|
2019-11-05 06:07:05 +08:00
|
|
|
import {debugRenderPhaseSideEffectsForStrictMode} from 'shared/ReactFeatureFlags';
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
|
2021-02-25 05:14:14 +08:00
|
|
|
import {StrictLegacyMode} from './ReactTypeOfMode';
|
2021-02-02 08:05:42 +08:00
|
|
|
import {
|
|
|
|
|
markSkippedUpdateLanes,
|
Refactor of interleaved ("concurrent") update queue (#24663)
* Always push updates to interleaved queue first
Interleaves updates (updates that are scheduled while another render
is already is progress) go into a special queue that isn't applied until
the end of the current render. They are transferred to the "real" queue
at the beginning of the next render.
Currently we check during `setState` whether an update should go
directly onto the real queue or onto the special interleaved queue. The
logic is subtle and it can lead to bugs if you mess it up, as in #24400.
Instead, this changes it to always go onto the interleaved queue. The
behavior is the same but the logic is simpler.
As a further step, we can also wait to update the `childLanes` until
the end of the current render. I'll do this in the next step.
* Move setState return path traversal to own call
A lot of the logic around scheduling an update needs access to the
fiber root. To obtain this reference, we must walk up the fiber return
path. We also do this to update `childLanes` on all the parent
nodes, so we can use the same traversal for both purposes.
The traversal currently happens inside `scheduleUpdateOnFiber`, but
sometimes we need to access it beyond that function, too.
So I've hoisted the traversal out of `scheduleUpdateOnFiber` into its
own function call that happens at the beginning of the
`setState` algorithm.
* Rename ReactInterleavedUpdates -> ReactFiberConcurrentUpdates
The scope of this module is expanding so I've renamed accordingly. No
behavioral changes.
* Enqueue and update childLanes in same function
During a setState, the childLanes are updated immediately, even if a
render is already in progress. This can lead to subtle concurrency bugs,
so the plan is to wait until the in-progress render has finished before
updating the childLanes, to prevent subtle concurrency bugs.
As a step toward that change, when scheduling an update, we should not
update the childLanes directly, but instead defer to the
ReactConcurrentUpdates module to do it at the appropriate time.
This makes markUpdateLaneFromFiberToRoot a private function that is
only called from the ReactConcurrentUpdates module.
* [FORKED] Don't update childLanes until after current render
(This is the riskiest commit in the stack. Only affects the "new"
reconciler fork.)
Updates that occur in a concurrent event while a render is already in
progress can't be processed during that render. This is tricky to get
right. Previously we solved this by adding concurrent updates to a
special `interleaved` queue, then transferring the `interleaved` queue
to the `pending` queue after the render phase had completed.
However, we would still mutate the `childLanes` along the parent path
immediately, which can lead to its own subtle data races.
Instead, we can queue the entire operation until after the render phase
has completed. This replaces the need for an `interleaved` field on
every fiber/hook queue.
The main motivation for this change, aside from simplifying the logic a
bit, is so we can read information about the current fiber while we're
walking up its return path, like whether it's inside a hidden tree.
(I haven't done anything like that in this commit, though.)
* Add 17691ac to forked revisions
2022-06-07 00:15:59 +08:00
|
|
|
isUnsafeClassRenderPhaseUpdate,
|
2022-06-30 11:31:42 +08:00
|
|
|
getWorkInProgressRootRenderLanes,
|
2022-12-02 12:19:13 +08:00
|
|
|
} from './ReactFiberWorkLoop';
|
Refactor of interleaved ("concurrent") update queue (#24663)
* Always push updates to interleaved queue first
Interleaves updates (updates that are scheduled while another render
is already is progress) go into a special queue that isn't applied until
the end of the current render. They are transferred to the "real" queue
at the beginning of the next render.
Currently we check during `setState` whether an update should go
directly onto the real queue or onto the special interleaved queue. The
logic is subtle and it can lead to bugs if you mess it up, as in #24400.
Instead, this changes it to always go onto the interleaved queue. The
behavior is the same but the logic is simpler.
As a further step, we can also wait to update the `childLanes` until
the end of the current render. I'll do this in the next step.
* Move setState return path traversal to own call
A lot of the logic around scheduling an update needs access to the
fiber root. To obtain this reference, we must walk up the fiber return
path. We also do this to update `childLanes` on all the parent
nodes, so we can use the same traversal for both purposes.
The traversal currently happens inside `scheduleUpdateOnFiber`, but
sometimes we need to access it beyond that function, too.
So I've hoisted the traversal out of `scheduleUpdateOnFiber` into its
own function call that happens at the beginning of the
`setState` algorithm.
* Rename ReactInterleavedUpdates -> ReactFiberConcurrentUpdates
The scope of this module is expanding so I've renamed accordingly. No
behavioral changes.
* Enqueue and update childLanes in same function
During a setState, the childLanes are updated immediately, even if a
render is already in progress. This can lead to subtle concurrency bugs,
so the plan is to wait until the in-progress render has finished before
updating the childLanes, to prevent subtle concurrency bugs.
As a step toward that change, when scheduling an update, we should not
update the childLanes directly, but instead defer to the
ReactConcurrentUpdates module to do it at the appropriate time.
This makes markUpdateLaneFromFiberToRoot a private function that is
only called from the ReactConcurrentUpdates module.
* [FORKED] Don't update childLanes until after current render
(This is the riskiest commit in the stack. Only affects the "new"
reconciler fork.)
Updates that occur in a concurrent event while a render is already in
progress can't be processed during that render. This is tricky to get
right. Previously we solved this by adding concurrent updates to a
special `interleaved` queue, then transferring the `interleaved` queue
to the `pending` queue after the render phase had completed.
However, we would still mutate the `childLanes` along the parent path
immediately, which can lead to its own subtle data races.
Instead, we can queue the entire operation until after the render phase
has completed. This replaces the need for an `interleaved` field on
every fiber/hook queue.
The main motivation for this change, aside from simplifying the logic a
bit, is so we can read information about the current fiber while we're
walking up its return path, like whether it's inside a hidden tree.
(I haven't done anything like that in this commit, though.)
* Add 17691ac to forked revisions
2022-06-07 00:15:59 +08:00
|
|
|
import {
|
|
|
|
|
enqueueConcurrentClassUpdate,
|
|
|
|
|
unsafe_markUpdateLaneFromFiberToRoot,
|
2022-12-02 12:19:13 +08:00
|
|
|
} from './ReactFiberConcurrentUpdates';
|
|
|
|
|
import {setIsStrictModeForDevtools} from './ReactFiberDevToolsHook';
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
|
2022-02-24 08:34:24 +08:00
|
|
|
import assign from 'shared/assign';
|
|
|
|
|
|
2022-09-10 04:03:48 +08:00
|
|
|
export type Update<State> = {
|
2020-06-12 11:05:15 +08:00
|
|
|
// TODO: Temporary field. Will remove this by storing a map of
|
|
|
|
|
// transition -> event time on the root.
|
|
|
|
|
eventTime: number,
|
|
|
|
|
lane: Lane,
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
|
|
|
|
|
tag: 0 | 1 | 2 | 3,
|
|
|
|
|
payload: any,
|
|
|
|
|
callback: (() => mixed) | null,
|
|
|
|
|
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
next: Update<State> | null,
|
2022-09-10 04:03:48 +08:00
|
|
|
};
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
|
2022-09-10 04:03:48 +08:00
|
|
|
export type SharedQueue<State> = {
|
2020-06-12 11:05:15 +08:00
|
|
|
pending: Update<State> | null,
|
2021-01-28 03:55:27 +08:00
|
|
|
lanes: Lanes,
|
2022-07-08 23:51:40 +08:00
|
|
|
hiddenCallbacks: Array<() => mixed> | null,
|
2022-09-10 04:03:48 +08:00
|
|
|
};
|
2019-12-11 08:42:42 +08:00
|
|
|
|
2022-09-10 04:03:48 +08:00
|
|
|
export type UpdateQueue<State> = {
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
baseState: State,
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
firstBaseUpdate: Update<State> | null,
|
|
|
|
|
lastBaseUpdate: Update<State> | null,
|
2019-12-11 08:42:42 +08:00
|
|
|
shared: SharedQueue<State>,
|
2022-07-08 23:51:40 +08:00
|
|
|
callbacks: Array<() => mixed> | null,
|
2022-09-10 04:03:48 +08:00
|
|
|
};
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
|
|
|
|
|
export const UpdateState = 0;
|
|
|
|
|
export const ReplaceState = 1;
|
|
|
|
|
export const ForceUpdate = 2;
|
|
|
|
|
export const CaptureUpdate = 3;
|
|
|
|
|
|
2018-05-15 10:18:47 +08:00
|
|
|
// Global state that is reset at the beginning of calling `processUpdateQueue`.
|
|
|
|
|
// It should only be read right after calling `processUpdateQueue`, via
|
|
|
|
|
// `checkHasForceUpdateAfterProcessing`.
|
|
|
|
|
let hasForceUpdate = false;
|
|
|
|
|
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
let didWarnUpdateInsideUpdate;
|
|
|
|
|
let currentlyProcessingQueue;
|
2022-09-14 05:57:38 +08:00
|
|
|
export let resetCurrentlyProcessingQueue: () => void;
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
if (__DEV__) {
|
|
|
|
|
didWarnUpdateInsideUpdate = false;
|
|
|
|
|
currentlyProcessingQueue = null;
|
|
|
|
|
resetCurrentlyProcessingQueue = () => {
|
|
|
|
|
currentlyProcessingQueue = null;
|
|
|
|
|
};
|
|
|
|
|
}
|
|
|
|
|
|
2019-12-11 08:42:42 +08:00
|
|
|
export function initializeUpdateQueue<State>(fiber: Fiber): void {
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
const queue: UpdateQueue<State> = {
|
2019-12-11 08:42:42 +08:00
|
|
|
baseState: fiber.memoizedState,
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
firstBaseUpdate: null,
|
|
|
|
|
lastBaseUpdate: null,
|
2019-12-11 08:42:42 +08:00
|
|
|
shared: {
|
|
|
|
|
pending: null,
|
2021-01-28 03:55:27 +08:00
|
|
|
lanes: NoLanes,
|
2022-07-08 23:51:40 +08:00
|
|
|
hiddenCallbacks: null,
|
2019-12-11 08:42:42 +08:00
|
|
|
},
|
2022-07-08 23:51:40 +08:00
|
|
|
callbacks: null,
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
};
|
2019-12-11 08:42:42 +08:00
|
|
|
fiber.updateQueue = queue;
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
}
|
|
|
|
|
|
2019-12-11 08:42:42 +08:00
|
|
|
export function cloneUpdateQueue<State>(
|
|
|
|
|
current: Fiber,
|
|
|
|
|
workInProgress: Fiber,
|
|
|
|
|
): void {
|
|
|
|
|
// Clone the update queue from current. Unless it's already a clone.
|
|
|
|
|
const queue: UpdateQueue<State> = (workInProgress.updateQueue: any);
|
|
|
|
|
const currentQueue: UpdateQueue<State> = (current.updateQueue: any);
|
|
|
|
|
if (queue === currentQueue) {
|
|
|
|
|
const clone: UpdateQueue<State> = {
|
|
|
|
|
baseState: currentQueue.baseState,
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
firstBaseUpdate: currentQueue.firstBaseUpdate,
|
|
|
|
|
lastBaseUpdate: currentQueue.lastBaseUpdate,
|
2019-12-11 08:42:42 +08:00
|
|
|
shared: currentQueue.shared,
|
2022-07-08 23:51:40 +08:00
|
|
|
callbacks: null,
|
2019-12-11 08:42:42 +08:00
|
|
|
};
|
|
|
|
|
workInProgress.updateQueue = clone;
|
|
|
|
|
}
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
}
|
|
|
|
|
|
2022-10-05 01:25:16 +08:00
|
|
|
export function createUpdate(eventTime: number, lane: Lane): Update<mixed> {
|
|
|
|
|
const update: Update<mixed> = {
|
2020-06-12 11:05:15 +08:00
|
|
|
eventTime,
|
|
|
|
|
lane,
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
|
|
|
|
|
tag: UpdateState,
|
|
|
|
|
payload: null,
|
|
|
|
|
callback: null,
|
|
|
|
|
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
next: null,
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
};
|
2019-07-23 05:17:43 +08:00
|
|
|
return update;
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
}
|
|
|
|
|
|
2021-02-02 08:05:42 +08:00
|
|
|
export function enqueueUpdate<State>(
|
|
|
|
|
fiber: Fiber,
|
|
|
|
|
update: Update<State>,
|
|
|
|
|
lane: Lane,
|
Refactor of interleaved ("concurrent") update queue (#24663)
* Always push updates to interleaved queue first
Interleaves updates (updates that are scheduled while another render
is already is progress) go into a special queue that isn't applied until
the end of the current render. They are transferred to the "real" queue
at the beginning of the next render.
Currently we check during `setState` whether an update should go
directly onto the real queue or onto the special interleaved queue. The
logic is subtle and it can lead to bugs if you mess it up, as in #24400.
Instead, this changes it to always go onto the interleaved queue. The
behavior is the same but the logic is simpler.
As a further step, we can also wait to update the `childLanes` until
the end of the current render. I'll do this in the next step.
* Move setState return path traversal to own call
A lot of the logic around scheduling an update needs access to the
fiber root. To obtain this reference, we must walk up the fiber return
path. We also do this to update `childLanes` on all the parent
nodes, so we can use the same traversal for both purposes.
The traversal currently happens inside `scheduleUpdateOnFiber`, but
sometimes we need to access it beyond that function, too.
So I've hoisted the traversal out of `scheduleUpdateOnFiber` into its
own function call that happens at the beginning of the
`setState` algorithm.
* Rename ReactInterleavedUpdates -> ReactFiberConcurrentUpdates
The scope of this module is expanding so I've renamed accordingly. No
behavioral changes.
* Enqueue and update childLanes in same function
During a setState, the childLanes are updated immediately, even if a
render is already in progress. This can lead to subtle concurrency bugs,
so the plan is to wait until the in-progress render has finished before
updating the childLanes, to prevent subtle concurrency bugs.
As a step toward that change, when scheduling an update, we should not
update the childLanes directly, but instead defer to the
ReactConcurrentUpdates module to do it at the appropriate time.
This makes markUpdateLaneFromFiberToRoot a private function that is
only called from the ReactConcurrentUpdates module.
* [FORKED] Don't update childLanes until after current render
(This is the riskiest commit in the stack. Only affects the "new"
reconciler fork.)
Updates that occur in a concurrent event while a render is already in
progress can't be processed during that render. This is tricky to get
right. Previously we solved this by adding concurrent updates to a
special `interleaved` queue, then transferring the `interleaved` queue
to the `pending` queue after the render phase had completed.
However, we would still mutate the `childLanes` along the parent path
immediately, which can lead to its own subtle data races.
Instead, we can queue the entire operation until after the render phase
has completed. This replaces the need for an `interleaved` field on
every fiber/hook queue.
The main motivation for this change, aside from simplifying the logic a
bit, is so we can read information about the current fiber while we're
walking up its return path, like whether it's inside a hidden tree.
(I haven't done anything like that in this commit, though.)
* Add 17691ac to forked revisions
2022-06-07 00:15:59 +08:00
|
|
|
): FiberRoot | null {
|
2019-12-11 08:42:42 +08:00
|
|
|
const updateQueue = fiber.updateQueue;
|
|
|
|
|
if (updateQueue === null) {
|
|
|
|
|
// Only occurs if the fiber has been unmounted.
|
Refactor of interleaved ("concurrent") update queue (#24663)
* Always push updates to interleaved queue first
Interleaves updates (updates that are scheduled while another render
is already is progress) go into a special queue that isn't applied until
the end of the current render. They are transferred to the "real" queue
at the beginning of the next render.
Currently we check during `setState` whether an update should go
directly onto the real queue or onto the special interleaved queue. The
logic is subtle and it can lead to bugs if you mess it up, as in #24400.
Instead, this changes it to always go onto the interleaved queue. The
behavior is the same but the logic is simpler.
As a further step, we can also wait to update the `childLanes` until
the end of the current render. I'll do this in the next step.
* Move setState return path traversal to own call
A lot of the logic around scheduling an update needs access to the
fiber root. To obtain this reference, we must walk up the fiber return
path. We also do this to update `childLanes` on all the parent
nodes, so we can use the same traversal for both purposes.
The traversal currently happens inside `scheduleUpdateOnFiber`, but
sometimes we need to access it beyond that function, too.
So I've hoisted the traversal out of `scheduleUpdateOnFiber` into its
own function call that happens at the beginning of the
`setState` algorithm.
* Rename ReactInterleavedUpdates -> ReactFiberConcurrentUpdates
The scope of this module is expanding so I've renamed accordingly. No
behavioral changes.
* Enqueue and update childLanes in same function
During a setState, the childLanes are updated immediately, even if a
render is already in progress. This can lead to subtle concurrency bugs,
so the plan is to wait until the in-progress render has finished before
updating the childLanes, to prevent subtle concurrency bugs.
As a step toward that change, when scheduling an update, we should not
update the childLanes directly, but instead defer to the
ReactConcurrentUpdates module to do it at the appropriate time.
This makes markUpdateLaneFromFiberToRoot a private function that is
only called from the ReactConcurrentUpdates module.
* [FORKED] Don't update childLanes until after current render
(This is the riskiest commit in the stack. Only affects the "new"
reconciler fork.)
Updates that occur in a concurrent event while a render is already in
progress can't be processed during that render. This is tricky to get
right. Previously we solved this by adding concurrent updates to a
special `interleaved` queue, then transferring the `interleaved` queue
to the `pending` queue after the render phase had completed.
However, we would still mutate the `childLanes` along the parent path
immediately, which can lead to its own subtle data races.
Instead, we can queue the entire operation until after the render phase
has completed. This replaces the need for an `interleaved` field on
every fiber/hook queue.
The main motivation for this change, aside from simplifying the logic a
bit, is so we can read information about the current fiber while we're
walking up its return path, like whether it's inside a hidden tree.
(I haven't done anything like that in this commit, though.)
* Add 17691ac to forked revisions
2022-06-07 00:15:59 +08:00
|
|
|
return null;
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
}
|
2019-12-11 08:42:42 +08:00
|
|
|
|
Split cross-package types from implementation
Some of our internal reconciler types have leaked into other packages.
Usually, these types are treated as opaque; we don't read and write
to its fields. This is good.
However, the type is often passed back to a reconciler method. For
example, React DOM creates a FiberRoot with `createContainer`, then
passes that root to `updateContainer`. It doesn't do anything with the
root except pass it through, but because `updateContainer` expects a
full FiberRoot, React DOM is still coupled to all its fields.
I don't know if there's an idiomatic way to handle this in Flow. Opaque
types are simlar, but those only work within a single file. AFAIK,
there's no way to use a package as the boundary for opaqueness.
The immediate problem this presents is that the reconciler refactor will
involve changes to our internal data structures. I don't want to have to
fork every single package that happens to pass through a Fiber or
FiberRoot, or access any one of its fields. So my current plan is to
share the same Flow type across both forks. The shared type will be a
superset of each implementation's type, e.g. Fiber will have both an
`expirationTime` field and a `lanes` field. The implementations will
diverge, but not the types.
To do this, I lifted the type definitions into a separate module.
2020-04-09 14:48:24 +08:00
|
|
|
const sharedQueue: SharedQueue<State> = (updateQueue: any).shared;
|
2021-02-02 08:05:42 +08:00
|
|
|
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
if (__DEV__) {
|
|
|
|
|
if (
|
2019-12-11 08:42:42 +08:00
|
|
|
currentlyProcessingQueue === sharedQueue &&
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
!didWarnUpdateInsideUpdate
|
|
|
|
|
) {
|
2022-12-03 01:14:12 +08:00
|
|
|
const componentName = getComponentNameFromFiber(fiber);
|
2019-12-15 02:09:25 +08:00
|
|
|
console.error(
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
'An update (setState, replaceState, or forceUpdate) was scheduled ' +
|
|
|
|
|
'from inside an update function. Update functions should be pure, ' +
|
|
|
|
|
'with zero side-effects. Consider using componentDidUpdate or a ' +
|
2022-12-03 01:14:12 +08:00
|
|
|
'callback.\n\nPlease update the following component: %s',
|
|
|
|
|
componentName,
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
);
|
|
|
|
|
didWarnUpdateInsideUpdate = true;
|
|
|
|
|
}
|
|
|
|
|
}
|
Refactor of interleaved ("concurrent") update queue (#24663)
* Always push updates to interleaved queue first
Interleaves updates (updates that are scheduled while another render
is already is progress) go into a special queue that isn't applied until
the end of the current render. They are transferred to the "real" queue
at the beginning of the next render.
Currently we check during `setState` whether an update should go
directly onto the real queue or onto the special interleaved queue. The
logic is subtle and it can lead to bugs if you mess it up, as in #24400.
Instead, this changes it to always go onto the interleaved queue. The
behavior is the same but the logic is simpler.
As a further step, we can also wait to update the `childLanes` until
the end of the current render. I'll do this in the next step.
* Move setState return path traversal to own call
A lot of the logic around scheduling an update needs access to the
fiber root. To obtain this reference, we must walk up the fiber return
path. We also do this to update `childLanes` on all the parent
nodes, so we can use the same traversal for both purposes.
The traversal currently happens inside `scheduleUpdateOnFiber`, but
sometimes we need to access it beyond that function, too.
So I've hoisted the traversal out of `scheduleUpdateOnFiber` into its
own function call that happens at the beginning of the
`setState` algorithm.
* Rename ReactInterleavedUpdates -> ReactFiberConcurrentUpdates
The scope of this module is expanding so I've renamed accordingly. No
behavioral changes.
* Enqueue and update childLanes in same function
During a setState, the childLanes are updated immediately, even if a
render is already in progress. This can lead to subtle concurrency bugs,
so the plan is to wait until the in-progress render has finished before
updating the childLanes, to prevent subtle concurrency bugs.
As a step toward that change, when scheduling an update, we should not
update the childLanes directly, but instead defer to the
ReactConcurrentUpdates module to do it at the appropriate time.
This makes markUpdateLaneFromFiberToRoot a private function that is
only called from the ReactConcurrentUpdates module.
* [FORKED] Don't update childLanes until after current render
(This is the riskiest commit in the stack. Only affects the "new"
reconciler fork.)
Updates that occur in a concurrent event while a render is already in
progress can't be processed during that render. This is tricky to get
right. Previously we solved this by adding concurrent updates to a
special `interleaved` queue, then transferring the `interleaved` queue
to the `pending` queue after the render phase had completed.
However, we would still mutate the `childLanes` along the parent path
immediately, which can lead to its own subtle data races.
Instead, we can queue the entire operation until after the render phase
has completed. This replaces the need for an `interleaved` field on
every fiber/hook queue.
The main motivation for this change, aside from simplifying the logic a
bit, is so we can read information about the current fiber while we're
walking up its return path, like whether it's inside a hidden tree.
(I haven't done anything like that in this commit, though.)
* Add 17691ac to forked revisions
2022-06-07 00:15:59 +08:00
|
|
|
|
|
|
|
|
if (isUnsafeClassRenderPhaseUpdate(fiber)) {
|
|
|
|
|
// This is an unsafe render phase update. Add directly to the update
|
|
|
|
|
// queue so we can process it immediately during the current render.
|
|
|
|
|
const pending = sharedQueue.pending;
|
|
|
|
|
if (pending === null) {
|
|
|
|
|
// This is the first update. Create a circular list.
|
|
|
|
|
update.next = update;
|
|
|
|
|
} else {
|
|
|
|
|
update.next = pending.next;
|
|
|
|
|
pending.next = update;
|
|
|
|
|
}
|
|
|
|
|
sharedQueue.pending = update;
|
|
|
|
|
|
|
|
|
|
// Update the childLanes even though we're most likely already rendering
|
|
|
|
|
// this fiber. This is for backwards compatibility in the case where you
|
|
|
|
|
// update a different component during render phase than the one that is
|
|
|
|
|
// currently renderings (a pattern that is accompanied by a warning).
|
|
|
|
|
return unsafe_markUpdateLaneFromFiberToRoot(fiber, lane);
|
|
|
|
|
} else {
|
|
|
|
|
return enqueueConcurrentClassUpdate(fiber, sharedQueue, update, lane);
|
|
|
|
|
}
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
}
|
|
|
|
|
|
2021-01-28 03:55:27 +08:00
|
|
|
export function entangleTransitions(root: FiberRoot, fiber: Fiber, lane: Lane) {
|
|
|
|
|
const updateQueue = fiber.updateQueue;
|
|
|
|
|
if (updateQueue === null) {
|
|
|
|
|
// Only occurs if the fiber has been unmounted.
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
const sharedQueue: SharedQueue<mixed> = (updateQueue: any).shared;
|
|
|
|
|
if (isTransitionLane(lane)) {
|
|
|
|
|
let queueLanes = sharedQueue.lanes;
|
|
|
|
|
|
|
|
|
|
// If any entangled lanes are no longer pending on the root, then they must
|
|
|
|
|
// have finished. We can remove them from the shared queue, which represents
|
|
|
|
|
// a superset of the actually pending lanes. In some cases we may entangle
|
|
|
|
|
// more than we need to, but that's OK. In fact it's worse if we *don't*
|
|
|
|
|
// entangle when we should.
|
|
|
|
|
queueLanes = intersectLanes(queueLanes, root.pendingLanes);
|
|
|
|
|
|
|
|
|
|
// Entangle the new transition lane with the other transition lanes.
|
|
|
|
|
const newQueueLanes = mergeLanes(queueLanes, lane);
|
2021-02-09 16:03:07 +08:00
|
|
|
sharedQueue.lanes = newQueueLanes;
|
|
|
|
|
// Even if queue.lanes already include lane, we don't know for certain if
|
|
|
|
|
// the lane finished since the last time we entangled it. So we need to
|
|
|
|
|
// entangle it again, just to be sure.
|
|
|
|
|
markRootEntangled(root, newQueueLanes);
|
2021-01-28 03:55:27 +08:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
export function enqueueCapturedUpdate<State>(
|
|
|
|
|
workInProgress: Fiber,
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
capturedUpdate: Update<State>,
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
) {
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
// Captured updates are updates that are thrown by a child during the render
|
|
|
|
|
// phase. They should be discarded if the render is aborted. Therefore,
|
|
|
|
|
// we should only put them on the work-in-progress queue, not the current one.
|
|
|
|
|
let queue: UpdateQueue<State> = (workInProgress.updateQueue: any);
|
|
|
|
|
|
|
|
|
|
// Check if the work-in-progress queue is a clone.
|
2019-12-11 08:42:42 +08:00
|
|
|
const current = workInProgress.alternate;
|
|
|
|
|
if (current !== null) {
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
const currentQueue: UpdateQueue<State> = (current.updateQueue: any);
|
|
|
|
|
if (queue === currentQueue) {
|
|
|
|
|
// The work-in-progress queue is the same as current. This happens when
|
|
|
|
|
// we bail out on a parent fiber that then captures an error thrown by
|
|
|
|
|
// a child. Since we want to append the update only to the work-in
|
|
|
|
|
// -progress queue, we need to clone the updates. We usually clone during
|
|
|
|
|
// processUpdateQueue, but that didn't happen in this case because we
|
|
|
|
|
// skipped over the parent when we bailed out.
|
|
|
|
|
let newFirst = null;
|
|
|
|
|
let newLast = null;
|
|
|
|
|
const firstBaseUpdate = queue.firstBaseUpdate;
|
|
|
|
|
if (firstBaseUpdate !== null) {
|
|
|
|
|
// Loop through the updates and clone them.
|
2022-10-05 03:39:26 +08:00
|
|
|
let update: Update<State> = firstBaseUpdate;
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
do {
|
|
|
|
|
const clone: Update<State> = {
|
2020-06-12 11:05:15 +08:00
|
|
|
eventTime: update.eventTime,
|
|
|
|
|
lane: update.lane,
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
|
|
|
|
|
tag: update.tag,
|
|
|
|
|
payload: update.payload,
|
2022-07-08 23:51:40 +08:00
|
|
|
// When this update is rebased, we should not fire its
|
|
|
|
|
// callback again.
|
|
|
|
|
callback: null,
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
|
|
|
|
|
next: null,
|
|
|
|
|
};
|
|
|
|
|
if (newLast === null) {
|
|
|
|
|
newFirst = newLast = clone;
|
|
|
|
|
} else {
|
|
|
|
|
newLast.next = clone;
|
|
|
|
|
newLast = clone;
|
|
|
|
|
}
|
2022-10-05 03:39:26 +08:00
|
|
|
// $FlowFixMe[incompatible-type] we bail out when we get a null
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
update = update.next;
|
|
|
|
|
} while (update !== null);
|
|
|
|
|
|
|
|
|
|
// Append the captured update the end of the cloned list.
|
|
|
|
|
if (newLast === null) {
|
|
|
|
|
newFirst = newLast = capturedUpdate;
|
|
|
|
|
} else {
|
|
|
|
|
newLast.next = capturedUpdate;
|
|
|
|
|
newLast = capturedUpdate;
|
|
|
|
|
}
|
|
|
|
|
} else {
|
|
|
|
|
// There are no base updates.
|
|
|
|
|
newFirst = newLast = capturedUpdate;
|
|
|
|
|
}
|
|
|
|
|
queue = {
|
|
|
|
|
baseState: currentQueue.baseState,
|
|
|
|
|
firstBaseUpdate: newFirst,
|
|
|
|
|
lastBaseUpdate: newLast,
|
|
|
|
|
shared: currentQueue.shared,
|
2022-07-08 23:51:40 +08:00
|
|
|
callbacks: currentQueue.callbacks,
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
};
|
|
|
|
|
workInProgress.updateQueue = queue;
|
|
|
|
|
return;
|
|
|
|
|
}
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Append the update to the end of the list.
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
const lastBaseUpdate = queue.lastBaseUpdate;
|
|
|
|
|
if (lastBaseUpdate === null) {
|
|
|
|
|
queue.firstBaseUpdate = capturedUpdate;
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
} else {
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
lastBaseUpdate.next = capturedUpdate;
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
}
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
queue.lastBaseUpdate = capturedUpdate;
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
function getStateFromUpdate<State>(
|
|
|
|
|
workInProgress: Fiber,
|
|
|
|
|
queue: UpdateQueue<State>,
|
|
|
|
|
update: Update<State>,
|
|
|
|
|
prevState: State,
|
|
|
|
|
nextProps: any,
|
|
|
|
|
instance: any,
|
|
|
|
|
): any {
|
|
|
|
|
switch (update.tag) {
|
|
|
|
|
case ReplaceState: {
|
|
|
|
|
const payload = update.payload;
|
|
|
|
|
if (typeof payload === 'function') {
|
|
|
|
|
// Updater function
|
|
|
|
|
if (__DEV__) {
|
2019-01-25 03:31:20 +08:00
|
|
|
enterDisallowedContextReadInDEV();
|
2020-04-09 11:54:54 +08:00
|
|
|
}
|
|
|
|
|
const nextState = payload.call(instance, prevState, nextProps);
|
|
|
|
|
if (__DEV__) {
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
if (
|
2019-11-05 06:07:05 +08:00
|
|
|
debugRenderPhaseSideEffectsForStrictMode &&
|
2021-02-25 05:14:14 +08:00
|
|
|
workInProgress.mode & StrictLegacyMode
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
) {
|
2021-08-26 06:35:38 +08:00
|
|
|
setIsStrictModeForDevtools(true);
|
2020-04-09 07:43:51 +08:00
|
|
|
try {
|
|
|
|
|
payload.call(instance, prevState, nextProps);
|
|
|
|
|
} finally {
|
2021-08-26 06:35:38 +08:00
|
|
|
setIsStrictModeForDevtools(false);
|
2020-04-09 07:43:51 +08:00
|
|
|
}
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
}
|
2019-01-25 03:31:20 +08:00
|
|
|
exitDisallowedContextReadInDEV();
|
|
|
|
|
}
|
|
|
|
|
return nextState;
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
}
|
|
|
|
|
// State object
|
|
|
|
|
return payload;
|
|
|
|
|
}
|
|
|
|
|
case CaptureUpdate: {
|
2020-09-05 05:34:07 +08:00
|
|
|
workInProgress.flags =
|
|
|
|
|
(workInProgress.flags & ~ShouldCapture) | DidCapture;
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
}
|
|
|
|
|
// Intentional fallthrough
|
|
|
|
|
case UpdateState: {
|
|
|
|
|
const payload = update.payload;
|
|
|
|
|
let partialState;
|
|
|
|
|
if (typeof payload === 'function') {
|
|
|
|
|
// Updater function
|
|
|
|
|
if (__DEV__) {
|
2019-01-25 03:31:20 +08:00
|
|
|
enterDisallowedContextReadInDEV();
|
2020-04-09 11:54:54 +08:00
|
|
|
}
|
|
|
|
|
partialState = payload.call(instance, prevState, nextProps);
|
|
|
|
|
if (__DEV__) {
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
if (
|
2019-11-05 06:07:05 +08:00
|
|
|
debugRenderPhaseSideEffectsForStrictMode &&
|
2021-02-25 05:14:14 +08:00
|
|
|
workInProgress.mode & StrictLegacyMode
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
) {
|
2021-08-26 06:35:38 +08:00
|
|
|
setIsStrictModeForDevtools(true);
|
2020-04-09 07:43:51 +08:00
|
|
|
try {
|
|
|
|
|
payload.call(instance, prevState, nextProps);
|
|
|
|
|
} finally {
|
2021-08-26 06:35:38 +08:00
|
|
|
setIsStrictModeForDevtools(false);
|
2020-04-09 07:43:51 +08:00
|
|
|
}
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
}
|
2019-01-25 03:31:20 +08:00
|
|
|
exitDisallowedContextReadInDEV();
|
|
|
|
|
}
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
} else {
|
|
|
|
|
// Partial state object
|
|
|
|
|
partialState = payload;
|
|
|
|
|
}
|
|
|
|
|
if (partialState === null || partialState === undefined) {
|
|
|
|
|
// Null and undefined are treated as no-ops.
|
|
|
|
|
return prevState;
|
|
|
|
|
}
|
|
|
|
|
// Merge the partial state and the previous state.
|
2022-02-24 08:34:24 +08:00
|
|
|
return assign({}, prevState, partialState);
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
}
|
|
|
|
|
case ForceUpdate: {
|
2018-05-15 10:18:47 +08:00
|
|
|
hasForceUpdate = true;
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
return prevState;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
return prevState;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
export function processUpdateQueue<State>(
|
|
|
|
|
workInProgress: Fiber,
|
|
|
|
|
props: any,
|
|
|
|
|
instance: any,
|
2020-06-12 11:05:15 +08:00
|
|
|
renderLanes: Lanes,
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
): void {
|
2019-12-11 08:42:42 +08:00
|
|
|
// This is always non-null on a ClassComponent or HostRoot
|
|
|
|
|
const queue: UpdateQueue<State> = (workInProgress.updateQueue: any);
|
2018-05-15 10:18:47 +08:00
|
|
|
|
2019-12-11 08:42:42 +08:00
|
|
|
hasForceUpdate = false;
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
|
|
|
|
|
if (__DEV__) {
|
2022-09-14 01:33:43 +08:00
|
|
|
// $FlowFixMe[escaped-generic] discovered when updating Flow
|
2019-12-11 08:42:42 +08:00
|
|
|
currentlyProcessingQueue = queue.shared;
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
}
|
|
|
|
|
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
let firstBaseUpdate = queue.firstBaseUpdate;
|
|
|
|
|
let lastBaseUpdate = queue.lastBaseUpdate;
|
2019-12-11 08:42:42 +08:00
|
|
|
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
// Check if there are pending updates. If so, transfer them to the base queue.
|
2019-12-11 08:42:42 +08:00
|
|
|
let pendingQueue = queue.shared.pending;
|
|
|
|
|
if (pendingQueue !== null) {
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
queue.shared.pending = null;
|
2019-12-11 08:42:42 +08:00
|
|
|
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
// The pending queue is circular. Disconnect the pointer between first
|
|
|
|
|
// and last so that it's non-circular.
|
|
|
|
|
const lastPendingUpdate = pendingQueue;
|
|
|
|
|
const firstPendingUpdate = lastPendingUpdate.next;
|
|
|
|
|
lastPendingUpdate.next = null;
|
|
|
|
|
// Append pending updates to base queue
|
|
|
|
|
if (lastBaseUpdate === null) {
|
|
|
|
|
firstBaseUpdate = firstPendingUpdate;
|
|
|
|
|
} else {
|
|
|
|
|
lastBaseUpdate.next = firstPendingUpdate;
|
|
|
|
|
}
|
|
|
|
|
lastBaseUpdate = lastPendingUpdate;
|
2019-12-11 08:42:42 +08:00
|
|
|
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
// If there's a current queue, and it's different from the base queue, then
|
|
|
|
|
// we need to transfer the updates to that queue, too. Because the base
|
|
|
|
|
// queue is a singly-linked list with no cycles, we can append to both
|
|
|
|
|
// lists and take advantage of structural sharing.
|
2019-12-11 08:42:42 +08:00
|
|
|
// TODO: Pass `current` as argument
|
|
|
|
|
const current = workInProgress.alternate;
|
|
|
|
|
if (current !== null) {
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
// This is always non-null on a ClassComponent or HostRoot
|
|
|
|
|
const currentQueue: UpdateQueue<State> = (current.updateQueue: any);
|
|
|
|
|
const currentLastBaseUpdate = currentQueue.lastBaseUpdate;
|
|
|
|
|
if (currentLastBaseUpdate !== lastBaseUpdate) {
|
|
|
|
|
if (currentLastBaseUpdate === null) {
|
|
|
|
|
currentQueue.firstBaseUpdate = firstPendingUpdate;
|
|
|
|
|
} else {
|
|
|
|
|
currentLastBaseUpdate.next = firstPendingUpdate;
|
|
|
|
|
}
|
|
|
|
|
currentQueue.lastBaseUpdate = lastPendingUpdate;
|
2019-12-10 05:19:12 +08:00
|
|
|
}
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
}
|
2019-12-10 05:19:12 +08:00
|
|
|
}
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
|
2019-12-11 08:42:42 +08:00
|
|
|
// These values may change as we process the queue.
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
if (firstBaseUpdate !== null) {
|
2019-12-11 08:42:42 +08:00
|
|
|
// Iterate through the list of updates to compute the result.
|
|
|
|
|
let newState = queue.baseState;
|
2020-06-12 11:05:15 +08:00
|
|
|
// TODO: Don't need to accumulate this. Instead, we can remove renderLanes
|
|
|
|
|
// from the original lanes.
|
|
|
|
|
let newLanes = NoLanes;
|
2019-12-11 08:42:42 +08:00
|
|
|
|
|
|
|
|
let newBaseState = null;
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
let newFirstBaseUpdate = null;
|
|
|
|
|
let newLastBaseUpdate = null;
|
|
|
|
|
|
2022-10-05 03:39:26 +08:00
|
|
|
let update: Update<State> = firstBaseUpdate;
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
do {
|
2022-06-30 11:31:42 +08:00
|
|
|
// TODO: Don't need this field anymore
|
2020-06-12 11:05:15 +08:00
|
|
|
const updateEventTime = update.eventTime;
|
2022-06-30 11:31:42 +08:00
|
|
|
|
|
|
|
|
// An extra OffscreenLane bit is added to updates that were made to
|
|
|
|
|
// a hidden tree, so that we can distinguish them from updates that were
|
|
|
|
|
// already there when the tree was hidden.
|
|
|
|
|
const updateLane = removeLanes(update.lane, OffscreenLane);
|
|
|
|
|
const isHiddenUpdate = updateLane !== update.lane;
|
|
|
|
|
|
|
|
|
|
// Check if this update was made while the tree was hidden. If so, then
|
|
|
|
|
// it's not a "base" update and we should disregard the extra base lanes
|
|
|
|
|
// that were added to renderLanes when we entered the Offscreen tree.
|
|
|
|
|
const shouldSkipUpdate = isHiddenUpdate
|
|
|
|
|
? !isSubsetOfLanes(getWorkInProgressRootRenderLanes(), updateLane)
|
|
|
|
|
: !isSubsetOfLanes(renderLanes, updateLane);
|
|
|
|
|
|
|
|
|
|
if (shouldSkipUpdate) {
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
// Priority is insufficient. Skip this update. If this is the first
|
|
|
|
|
// skipped update, the previous update/state is the new base
|
|
|
|
|
// update/state.
|
|
|
|
|
const clone: Update<State> = {
|
2020-06-12 11:05:15 +08:00
|
|
|
eventTime: updateEventTime,
|
|
|
|
|
lane: updateLane,
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
|
|
|
|
|
tag: update.tag,
|
|
|
|
|
payload: update.payload,
|
|
|
|
|
callback: update.callback,
|
|
|
|
|
|
|
|
|
|
next: null,
|
|
|
|
|
};
|
|
|
|
|
if (newLastBaseUpdate === null) {
|
|
|
|
|
newFirstBaseUpdate = newLastBaseUpdate = clone;
|
|
|
|
|
newBaseState = newState;
|
|
|
|
|
} else {
|
|
|
|
|
newLastBaseUpdate = newLastBaseUpdate.next = clone;
|
|
|
|
|
}
|
|
|
|
|
// Update the remaining priority in the queue.
|
2020-06-12 11:05:15 +08:00
|
|
|
newLanes = mergeLanes(newLanes, updateLane);
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
} else {
|
|
|
|
|
// This update does have sufficient priority.
|
|
|
|
|
|
|
|
|
|
if (newLastBaseUpdate !== null) {
|
2019-12-11 08:42:42 +08:00
|
|
|
const clone: Update<State> = {
|
2020-06-12 11:05:15 +08:00
|
|
|
eventTime: updateEventTime,
|
|
|
|
|
// This update is going to be committed so we never want uncommit
|
|
|
|
|
// it. Using NoLane works because 0 is a subset of all bitmasks, so
|
|
|
|
|
// this will never be skipped by the check above.
|
|
|
|
|
lane: NoLane,
|
2019-12-11 08:42:42 +08:00
|
|
|
|
|
|
|
|
tag: update.tag,
|
|
|
|
|
payload: update.payload,
|
2022-07-08 23:51:40 +08:00
|
|
|
|
|
|
|
|
// When this update is rebased, we should not fire its
|
|
|
|
|
// callback again.
|
|
|
|
|
callback: null,
|
2019-12-11 08:42:42 +08:00
|
|
|
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
next: null,
|
2019-12-11 08:42:42 +08:00
|
|
|
};
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
newLastBaseUpdate = newLastBaseUpdate.next = clone;
|
2019-12-10 05:19:12 +08:00
|
|
|
}
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
|
|
|
|
|
// Process this update.
|
|
|
|
|
newState = getStateFromUpdate(
|
|
|
|
|
workInProgress,
|
|
|
|
|
queue,
|
|
|
|
|
update,
|
|
|
|
|
newState,
|
|
|
|
|
props,
|
|
|
|
|
instance,
|
|
|
|
|
);
|
|
|
|
|
const callback = update.callback;
|
2022-07-08 23:51:40 +08:00
|
|
|
if (callback !== null) {
|
2020-09-05 05:34:07 +08:00
|
|
|
workInProgress.flags |= Callback;
|
2022-07-08 23:51:40 +08:00
|
|
|
if (isHiddenUpdate) {
|
|
|
|
|
workInProgress.flags |= Visibility;
|
|
|
|
|
}
|
|
|
|
|
const callbacks = queue.callbacks;
|
|
|
|
|
if (callbacks === null) {
|
|
|
|
|
queue.callbacks = [callback];
|
2019-12-11 08:42:42 +08:00
|
|
|
} else {
|
2022-07-08 23:51:40 +08:00
|
|
|
callbacks.push(callback);
|
2019-12-11 08:42:42 +08:00
|
|
|
}
|
|
|
|
|
}
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
}
|
2022-10-05 03:39:26 +08:00
|
|
|
// $FlowFixMe[incompatible-type] we bail out when we get a null
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
update = update.next;
|
|
|
|
|
if (update === null) {
|
|
|
|
|
pendingQueue = queue.shared.pending;
|
|
|
|
|
if (pendingQueue === null) {
|
|
|
|
|
break;
|
|
|
|
|
} else {
|
|
|
|
|
// An update was scheduled from inside a reducer. Add the new
|
|
|
|
|
// pending updates to the end of the list and keep processing.
|
|
|
|
|
const lastPendingUpdate = pendingQueue;
|
|
|
|
|
// Intentionally unsound. Pending updates form a circular list, but we
|
|
|
|
|
// unravel them when transferring them to the base queue.
|
2023-01-31 21:25:05 +08:00
|
|
|
const firstPendingUpdate =
|
|
|
|
|
((lastPendingUpdate.next: any): Update<State>);
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
lastPendingUpdate.next = null;
|
|
|
|
|
update = firstPendingUpdate;
|
|
|
|
|
queue.lastBaseUpdate = lastPendingUpdate;
|
|
|
|
|
queue.shared.pending = null;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
} while (true);
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
if (newLastBaseUpdate === null) {
|
2019-12-11 08:42:42 +08:00
|
|
|
newBaseState = newState;
|
|
|
|
|
}
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
|
2019-12-11 08:42:42 +08:00
|
|
|
queue.baseState = ((newBaseState: any): State);
|
Bugfix: "Captured" updates on legacy queue (#18265)
* Bugfix: "Captured" updates on legacy queue
This fixes a bug with error boundaries. Error boundaries have a notion
of "captured" updates that represent errors that are thrown in its
subtree during the render phase. These updates are meant to be dropped
if the render is aborted.
The bug happens when there's a concurrent update (an update from an
interleaved event) in between when the error is thrown and when the
error boundary does its second pass. The concurrent update is
transferred from the pending queue onto the base queue. Usually, at this
point the base queue is the same as the current queue. So when we
append the pending updates to the work-in-progress queue, it also
appends to the current queue.
However, in the case of an error boundary's second pass, the base queue
has already forked from the current queue; it includes both the
"captured" updates and any concurrent updates. In that case, what we
need to do is append separately to both queues. Which we weren't doing.
That isn't the full story, though. You would expect that this mistake
would manifest as dropping the interleaved updates. But instead what
was happening is that the "captured" updates, the ones that are meant
to be dropped if the render is aborted, were being added to the
current queue.
The reason is that the `baseQueue` structure is a circular linked list.
The motivation for this was to save memory; instead of separate `first`
and `last` pointers, you only need to point to `last`.
But this approach does not work with structural sharing. So what was
happening is that the captured updates were accidentally being added
to the current queue because of the circular link.
To fix this, I changed the `baseQueue` from a circular linked list to a
singly-linked list so that we can take advantage of structural sharing.
The "pending" queue, however, remains a circular list because it doesn't
need to be persistent.
This bug also affects the root fiber, which uses the same update queue
implementation and also acts like an error boundary.
It does not affect the hook update queue because they do not have any
notion of "captured" updates. So I've left it alone for now. However,
when we implement resuming, we will have to account for the same issue.
* Ensure base queue is a clone
When an error boundary captures an error, we append the error update
to the work-in-progress queue only so that if the render is aborted,
the error update is dropped.
Before appending to the queue, we need to make sure the queue is a
work-in-progress copy. Usually we clone the queue during
`processUpdateQueue`; however, if the base queue has lower priority
than the current render, we may have bailed out on the boundary fiber
without ever entering `processUpdateQueue`. So we need to lazily clone
the queue.
* Add warning to protect against refactor hazard
The hook queue does not have resuming or "captured" updates, but if
we ever add them in the future, we'll need to make sure we check if the
queue is forked before transfering the pending updates to them.
2020-03-12 02:53:45 +08:00
|
|
|
queue.firstBaseUpdate = newFirstBaseUpdate;
|
|
|
|
|
queue.lastBaseUpdate = newLastBaseUpdate;
|
2019-12-11 08:42:42 +08:00
|
|
|
|
2022-06-30 11:31:42 +08:00
|
|
|
if (firstBaseUpdate === null) {
|
2021-01-28 03:55:27 +08:00
|
|
|
// `queue.lanes` is used for entangling transitions. We can set it back to
|
|
|
|
|
// zero once the queue is empty.
|
|
|
|
|
queue.shared.lanes = NoLanes;
|
|
|
|
|
}
|
|
|
|
|
|
2019-12-11 08:42:42 +08:00
|
|
|
// Set the remaining expiration time to be whatever is remaining in the queue.
|
|
|
|
|
// This should be fine because the only two other things that contribute to
|
|
|
|
|
// expiration time are props and context. We're already in the middle of the
|
|
|
|
|
// begin phase by the time we start processing the queue, so we've already
|
|
|
|
|
// dealt with the props. Context in components that specify
|
|
|
|
|
// shouldComponentUpdate is tricky; but we'll have to account for
|
|
|
|
|
// that regardless.
|
2020-06-12 11:05:15 +08:00
|
|
|
markSkippedUpdateLanes(newLanes);
|
|
|
|
|
workInProgress.lanes = newLanes;
|
2019-12-11 08:42:42 +08:00
|
|
|
workInProgress.memoizedState = newState;
|
|
|
|
|
}
|
2019-12-10 05:19:12 +08:00
|
|
|
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
if (__DEV__) {
|
|
|
|
|
currentlyProcessingQueue = null;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2023-01-10 04:46:48 +08:00
|
|
|
function callCallback(callback: () => mixed, context: any) {
|
[RFC] Codemod invariant -> throw new Error (#22435)
* Hoist error codes import to module scope
When this code was written, the error codes map (`codes.json`) was
created on-the-fly, so we had to lazily require from inside the visitor.
Because `codes.json` is now checked into source, we can import it a
single time in module scope.
* Minify error constructors in production
We use a script to minify our error messages in production. Each message
is assigned an error code, defined in `scripts/error-codes/codes.json`.
Then our build script replaces the messages with a link to our
error decoder page, e.g. https://reactjs.org/docs/error-decoder.html/?invariant=92
This enables us to write helpful error messages without increasing the
bundle size.
Right now, the script only works for `invariant` calls. It does not work
if you throw an Error object. This is an old Facebookism that we don't
really need, other than the fact that our error minification script
relies on it.
So, I've updated the script to minify error constructors, too:
Input:
Error(`A ${adj} message that contains ${noun}`);
Output:
Error(formatProdErrorMessage(ERR_CODE, adj, noun));
It only works for constructors that are literally named Error, though we
could add support for other names, too.
As a next step, I will add a lint rule to enforce that errors written
this way must have a corresponding error code.
* Minify "no fallback UI specified" error in prod
This error message wasn't being minified because it doesn't use
invariant. The reason it didn't use invariant is because this particular
error is created without begin thrown — it doesn't need to be thrown
because it's located inside the error handling part of the runtime.
Now that the error minification script supports Error constructors, we
can minify it by assigning it a production error code in
`scripts/error-codes/codes.json`.
To support the use of Error constructors more generally, I will add a
lint rule that enforces each message has a corresponding error code.
* Lint rule to detect unminified errors
Adds a lint rule that detects when an Error constructor is used without
a corresponding production error code.
We already have this for `invariant`, but not for regular errors, i.e.
`throw new Error(msg)`. There's also nothing that enforces the use of
`invariant` besides convention.
There are some packages where we don't care to minify errors. These are
packages that run in environments where bundle size is not a concern,
like react-pg. I added an override in the ESLint config to ignore these.
* Temporarily add invariant codemod script
I'm adding this codemod to the repo temporarily, but I'll revert it
in the same PR. That way we don't have to check it in but it's still
accessible (via the PR) if we need it later.
* [Automated] Codemod invariant -> Error
This commit contains only automated changes:
npx jscodeshift -t scripts/codemod-invariant.js packages --ignore-pattern="node_modules/**/*"
yarn linc --fix
yarn prettier
I will do any manual touch ups in separate commits so they're easier
to review.
* Remove temporary codemod script
This reverts the codemod script and ESLint config I added temporarily
in order to perform the invariant codemod.
* Manual touch ups
A few manual changes I made after the codemod ran.
* Enable error code transform per package
Currently we're not consistent about which packages should have their
errors minified in production and which ones should.
This adds a field to the bundle configuration to control whether to
apply the transform. We should decide what the criteria is going
forward. I think it's probably a good idea to minify any package that
gets sent over the network. So yes to modules that run in the browser,
and no to modules that run on the server and during development only.
2021-10-01 03:01:28 +08:00
|
|
|
if (typeof callback !== 'function') {
|
|
|
|
|
throw new Error(
|
|
|
|
|
'Invalid argument passed as callback. Expected a function. Instead ' +
|
|
|
|
|
`received: ${callback}`,
|
|
|
|
|
);
|
|
|
|
|
}
|
|
|
|
|
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
callback.call(context);
|
|
|
|
|
}
|
|
|
|
|
|
2018-05-15 10:18:47 +08:00
|
|
|
export function resetHasForceUpdateBeforeProcessing() {
|
|
|
|
|
hasForceUpdate = false;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
export function checkHasForceUpdateAfterProcessing(): boolean {
|
|
|
|
|
return hasForceUpdate;
|
|
|
|
|
}
|
|
|
|
|
|
2022-07-08 23:51:40 +08:00
|
|
|
export function deferHiddenCallbacks<State>(
|
|
|
|
|
updateQueue: UpdateQueue<State>,
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
): void {
|
2022-07-08 23:51:40 +08:00
|
|
|
// When an update finishes on a hidden component, its callback should not
|
|
|
|
|
// be fired until/unless the component is made visible again. Stash the
|
|
|
|
|
// callback on the shared queue object so it can be fired later.
|
|
|
|
|
const newHiddenCallbacks = updateQueue.callbacks;
|
|
|
|
|
if (newHiddenCallbacks !== null) {
|
|
|
|
|
const existingHiddenCallbacks = updateQueue.shared.hiddenCallbacks;
|
|
|
|
|
if (existingHiddenCallbacks === null) {
|
|
|
|
|
updateQueue.shared.hiddenCallbacks = newHiddenCallbacks;
|
|
|
|
|
} else {
|
2023-01-31 21:25:05 +08:00
|
|
|
updateQueue.shared.hiddenCallbacks =
|
|
|
|
|
existingHiddenCallbacks.concat(newHiddenCallbacks);
|
2022-07-08 23:51:40 +08:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
export function commitHiddenCallbacks<State>(
|
|
|
|
|
updateQueue: UpdateQueue<State>,
|
|
|
|
|
context: any,
|
|
|
|
|
): void {
|
|
|
|
|
// This component is switching from hidden -> visible. Commit any callbacks
|
|
|
|
|
// that were previously deferred.
|
|
|
|
|
const hiddenCallbacks = updateQueue.shared.hiddenCallbacks;
|
|
|
|
|
if (hiddenCallbacks !== null) {
|
|
|
|
|
updateQueue.shared.hiddenCallbacks = null;
|
|
|
|
|
for (let i = 0; i < hiddenCallbacks.length; i++) {
|
|
|
|
|
const callback = hiddenCallbacks[i];
|
|
|
|
|
callCallback(callback, context);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
export function commitCallbacks<State>(
|
|
|
|
|
updateQueue: UpdateQueue<State>,
|
|
|
|
|
context: any,
|
|
|
|
|
): void {
|
|
|
|
|
const callbacks = updateQueue.callbacks;
|
|
|
|
|
if (callbacks !== null) {
|
|
|
|
|
updateQueue.callbacks = null;
|
|
|
|
|
for (let i = 0; i < callbacks.length; i++) {
|
|
|
|
|
const callback = callbacks[i];
|
|
|
|
|
callCallback(callback, context);
|
Decouple update queue from Fiber type (#12600)
* Decouple update queue from Fiber type
The update queue is in need of a refactor. Recent bugfixes (#12528) have
exposed some flaws in how it's modeled. Upcoming features like Suspense
and [redacted] also rely on the update queue in ways that weren't
anticipated in the original design.
Major changes:
- Instead of boolean flags for `isReplace` and `isForceUpdate`, updates
have a `tag` field (like Fiber). This lowers the cost for adding new
types of updates.
- Render phase updates are special cased. Updates scheduled during
the render phase are dropped if the work-in-progress does not commit.
This is used for `getDerivedStateFrom{Props,Catch}`.
- `callbackList` has been replaced with a generic effect list. Aside
from callbacks, this is also used for `componentDidCatch`.
* Remove first class UpdateQueue types and use closures instead
I tried to avoid this at first, since we avoid it everywhere else in the Fiber
codebase, but since updates are not in a hot path, the trade off with file size
seems worth it.
* Store captured errors on a separate part of the update queue
This way they can be reused independently of updates like
getDerivedStateFromProps. This will be important for resuming.
* Revert back to storing hasForceUpdate on the update queue
Instead of using the effect tag. Ideally, this would be part of the
return type of processUpdateQueue.
* Rename UpdateQueue effect type back to Callback
I don't love this name either, but it's less confusing than UpdateQueue
I suppose. Conceptually, this is usually a callback: setState callbacks,
componentDidCatch. The only case that feels a bit weird is Timeouts,
which use this effect to attach a promise listener. I guess that kinda
fits, too.
* Call getDerivedStateFromProps every render, even if props did not change
Rather than enqueue a new setState updater for every props change, we
can skip the update queue entirely and merge the result into state at
the end. This makes more sense, since "receiving props" is not an event
that should be observed. It's still a bit weird, since eventually we do
persist the derived state (in other words, it accumulates).
* Store captured effects on separate list from "own" effects (callbacks)
For resuming, we need the ability to discard the "own" effects while
reusing the captured effects.
* Optimize for class components
Change `process` and `callback` to match the expected payload types
for class components. I had intended for the update queue to be reusable
for both class components and a future React API, but we'll likely have
to fork anyway.
* Only double-invoke render phase lifecycles functions in DEV
* Use global state to track currently processing queue in DEV
2018-04-23 14:05:28 +08:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|