Commit Graph

16107 Commits

Author SHA1 Message Date
Daniel Dunbar 3967fca364 CMake: Add a clang-c++tests target for running the experimental C++ header
-fsyntax-only'ing tests.

llvm-svn: 82541
2009-09-22 10:07:55 +00:00
Daniel Dunbar 95d5a1d240 Add safe part of previous (reverted) commit, necessary to update to LLVM API change.
llvm-svn: 82540
2009-09-22 10:06:21 +00:00
Daniel Dunbar b879c3c125 Revert "Switch a few clients over to StringLiteral::getString.", this is breaking some projects, but I don't have a test case yet.
llvm-svn: 82539
2009-09-22 10:03:52 +00:00
Daniel Dunbar b45012dcde Update PCH serialization of FunctionDecl flags.
llvm-svn: 82526
2009-09-22 05:38:14 +00:00
Daniel Dunbar 2d925ebd0a Allow PCH files to be read from stdin.
llvm-svn: 82525
2009-09-22 05:38:01 +00:00
Daniel Dunbar b5cbf77c2e Remove old test runner, this has moved to LLVM/utils/lit and all known clients
have been updated.
 - Please let me know of any problems.

llvm-svn: 82524
2009-09-22 05:16:02 +00:00
Ted Kremenek f9539d0c3f Fix: <rdar://problem/7242015> [RegionStore] variable passed-by-reference (via integer) to function call not invalidated
llvm-svn: 82523
2009-09-22 04:48:39 +00:00
Daniel Dunbar 0512647362 Force triple in test.
llvm-svn: 82519
2009-09-22 03:50:39 +00:00
Ted Kremenek ece8524474 Update checker build.
llvm-svn: 82518
2009-09-22 03:38:38 +00:00
Daniel Dunbar 5de27daf93 Switch a few clients over to StringLiteral::getString.
- Switching all of them out-of-my-current-scope-of-interest, sorry.

llvm-svn: 82515
2009-09-22 03:27:52 +00:00
Daniel Dunbar 362178883c Add StringLiteral::getString -> StringRef.
llvm-svn: 82514
2009-09-22 03:27:33 +00:00
Anders Carlsson 8143069417 Store the set of indirect primary bases directly in the record layout builder.
llvm-svn: 82513
2009-09-22 03:02:06 +00:00
Mike Stump 207c680ff3 Fix some typos. WIP. Large alignments don't work yet.
llvm-svn: 82512
2009-09-22 02:44:17 +00:00
Mike Stump 8ce0ea1360 Fix doc.
llvm-svn: 82511
2009-09-22 02:43:44 +00:00
Daniel Dunbar a25a9dff0d Add test case for PR4894 (it was fixed in LLVM).
llvm-svn: 82510
2009-09-22 02:18:40 +00:00
Daniel Dunbar 1da62f5723 This FIXME is done.
llvm-svn: 82509
2009-09-22 02:17:27 +00:00
Mike Stump 2114d7c299 Improve debug info generation for __block variables.
llvm-svn: 82508
2009-09-22 02:12:52 +00:00
Anders Carlsson d6020c321a Record layout builder cleanup.
llvm-svn: 82502
2009-09-22 00:04:45 +00:00
John McCall 9dd450bb78 Change all the Type::getAsFoo() methods to specializations of Type::getAs().
Several of the existing methods were identical to their respective
specializations, and so have been removed entirely.  Several more 'leaf'
optimizations were introduced.

The getAsFoo() methods which imposed extra conditions, like
getAsObjCInterfacePointerType(), have been left in place.

llvm-svn: 82501
2009-09-21 23:43:11 +00:00
Ted Kremenek fd68c7bdc0 Add test case for <rdar://problem/6829164>, which was implicitly fixed in r79694.
llvm-svn: 82495
2009-09-21 23:22:11 +00:00
Fariborz Jahanian 71848a3ff1 Removed -fobjc-newgc-api option. clang now conforms to
gcc-style write-barrier api only.

llvm-svn: 82493
2009-09-21 23:03:37 +00:00
Ted Kremenek 25c9c1427a Provide intermediate solution to handling assignments to structs via an
integer pointer.  For now just invalidate the fields of the struct.

This addresses: <rdar://problem/7185607> [RegionStore] support invalidation of bit fields using integer assignment

llvm-svn: 82492
2009-09-21 22:58:52 +00:00
Douglas Gregor ac9b61375d Another code-completion TODO
llvm-svn: 82482
2009-09-21 20:52:01 +00:00
Douglas Gregor 9d64c5e3a5 Code completion for ordinary names when we're starting a declaration, expression, or statement
llvm-svn: 82481
2009-09-21 20:51:25 +00:00
Fariborz Jahanian a74039426d -fobjc-newgc-api is now the default.
llvm-svn: 82478
2009-09-21 20:17:37 +00:00
Douglas Gregor 2af2f670d1 When providing a code-completion suggestion for a hidden name, include
a nested-name-specifier that describes how to refer to that name. For
example, given:

  struct Base { int member; };
  struct Derived : Base { int member; };

the code-completion result for a member access into "Derived" will
provide both "member" to refer to Derived::member (no qualification needed) and
"Base::member" to refer to Base::member (qualification included).

llvm-svn: 82476
2009-09-21 20:12:40 +00:00
Douglas Gregor f251067530 Enhance "case" code completion in C++ to suggest qualified names for
enumerators when either the user intentionally wrote a qualified name
(in which case we just use that nested-name-specifier to match
the user's code) or when this is the first "case" statement and we
need a qualified name to refer to an enumerator in a different scope.

llvm-svn: 82474
2009-09-21 19:57:38 +00:00
Fariborz Jahanian 38c3ae9bc4 Miscellanous fixes in generatation of objc gc's write-barriers.
llvm-svn: 82472
2009-09-21 18:54:29 +00:00
Douglas Gregor d328d57c39 Code completion for "case" statements within a switch on an expression
of enumeration type, providing the various unused enumerators as options.

llvm-svn: 82467
2009-09-21 18:10:23 +00:00
Douglas Gregor 3545ff43f4 Refactor and simplify the CodeCompleteConsumer, so that all of the
real work is performed within Sema. Addresses Chris's comments, but
still retains the heavyweight list-of-multimaps data structure.

llvm-svn: 82459
2009-09-21 16:56:56 +00:00
Daniel Dunbar 607a2a1ed1 Add an XFAIL test which compiles differently from a .ast.
llvm-svn: 82437
2009-09-21 05:16:43 +00:00
Daniel Dunbar 143021e0ca Add missing PCH support for -fstack-protector.
llvm-svn: 82435
2009-09-21 04:16:19 +00:00
Chris Lattner bf20638966 Implement __builtin_unreachable(), a GCC 4.5 extension.
llvm-svn: 82433
2009-09-21 03:09:59 +00:00
Daniel Dunbar 26e37a77c6 Switch ProcessASTInputFile to still use ParseAST.
- Currently this requires us to fake an input file.

 - This allows Sema to be keep all the logic for how to pull decls out of the external AST source and how to handle things like tentative definitions.

llvm-svn: 82432
2009-09-21 03:03:56 +00:00
Daniel Dunbar b7bbfdd9e6 Change ASTUnit to only initialize the predefines buffer to the suggested predefines.
- It isn't really clear what to do with the preprocessor here, but this is more sensible.

llvm-svn: 82431
2009-09-21 03:03:47 +00:00
Daniel Dunbar 7cd285f0fe Change ASTUnit to take the Diagnostic as an argument, the client should have control of this.
llvm-svn: 82430
2009-09-21 03:03:39 +00:00
Daniel Dunbar 948062a592 Add Diagnostic to Indexer, and have it keep its own FileManager instead of taking an external reference (which was leaked in the case of the CIndex library).
llvm-svn: 82429
2009-09-21 03:03:22 +00:00
Anders Carlsson a18322cea5 Start mangling expressions.
llvm-svn: 82423
2009-09-21 01:21:10 +00:00
Daniel Dunbar 1d9b45304d Work around FileCheck -NOT restriction.
llvm-svn: 82417
2009-09-20 23:35:52 +00:00
Daniel Dunbar 22d5360ed0 Switch TestRunner.sh to just use 'lit'.
- Does people use this?

llvm-svn: 82406
2009-09-20 22:00:23 +00:00
Daniel Dunbar c6bca95723 Switch to using the new LLVM lit test runner by default.
llvm-svn: 82405
2009-09-20 22:00:17 +00:00
Daniel Dunbar a173d9b874 Include LLVM {src,obj} root in lit.site.cfg, to support non-standard layouts.
llvm-svn: 82401
2009-09-20 19:04:35 +00:00
Chris Lattner 78765c36de switch command line 'parse' methods to use StringRef for efficiency, which
is also required for an llvm-side change.

llvm-svn: 82344
2009-09-20 00:39:15 +00:00
Daniel Dunbar 9996138a80 Ok, an AssertingVH definitely doesn't work for now because we free our cache after the optimizer may have hacked on the module. Use a WeakVH instead.
llvm-svn: 82324
2009-09-19 20:17:48 +00:00
Nick Lewycky 41eaf0ac57 Make clang stop relying on ConstantStruct::get's default value for isPacked
which will be going away (ie. it's becoming a required parameter) later today.

llvm-svn: 82323
2009-09-19 20:00:52 +00:00
Daniel Dunbar 1cbaae56cf Switch CGDebugInfo type cache to using an AssertingVH.
llvm-svn: 82321
2009-09-19 19:27:24 +00:00
Daniel Dunbar de870bdcf2 Factor out CGDebugInfo::CreateTypeNode method.
- No functionality change.

llvm-svn: 82320
2009-09-19 19:27:14 +00:00
Benjamin Kramer dad401dbc1 Simplify code with a StringRef.
llvm-svn: 82308
2009-09-19 08:02:46 +00:00
Douglas Gregor b5c738b434 In C++ code completion, only suggest the "template" keyword after ".",
"->", or "::" if we will be looking into a dependent context. It's not
wrong to use the "template" keyword, but it's to needed, either.

llvm-svn: 82307
2009-09-18 23:55:56 +00:00
Douglas Gregor ce23bae4f8 Make the construction of the code-completion string for a function
template smarter, by taking into account which function template
parameters are deducible from the call arguments. For example,

  template<typename RandomAccessIterator>
  void sort(RandomAccessIterator first, RandomAccessIterator last);

will have a code-completion string like

  sort({RandomAccessIterator first}, {RandomAccessIterator last})

since the template argument for its template parameter is
deducible. On the other hand,

  template<class X, class Y>
  X* dyn_cast(Y *Val);

will have a code-completion string like

  dyn_cast<{class X}>({Y *Val})

since the template type parameter X is not deducible from the function
call.

llvm-svn: 82306
2009-09-18 23:21:38 +00:00