Commit Graph

325 Commits

Author SHA1 Message Date
Steve Naroff 6d40db0dde Implement a more sensible strategy for ObjC built-in types (addressing a long standing FIXME in Sema::GetObjcIdType()).
This removes several gross hacks to work around the previous "lazy" behavior.

Two notes:
- MinimalActions still needs to be taught about the built-in types (This breaks one of the -noop test cases). I started this, then added a FIXME.
- I didn't convert Sema::GetObjcProtoType() yet.

llvm-svn: 43567
2007-10-31 18:42:27 +00:00
Steve Naroff ac0216385f Checking in some code that is still under construction.
I need to (finally) change the way Class/id/SEL/IMP are built-in...the current approach of doing it in the preprocessor is "broken". The other problem is Sema::GetObjcIdType/GetObjcSelType/GetObjcClassType, the hooks that initialize ASTContext lazily. These built-in types need to be done up front...

llvm-svn: 43557
2007-10-31 16:03:04 +00:00
Steve Naroff 37e011ce43 Add some plumbing to help cope with rewriting "id<p>", "Class<p>*".
llvm-svn: 43543
2007-10-31 04:38:33 +00:00
Steve Naroff 5cdcd9b61c Stop pre-defining objc_msgSend/objc_getClass in the preprocessor. Instead, I generate these declaration on the fly when rewriting a message expression.
llvm-svn: 43529
2007-10-30 23:14:51 +00:00
Chris Lattner 39afd571a0 Avoid using macro expansions, which causes these to come out in -E mode.
llvm-svn: 43505
2007-10-30 17:45:43 +00:00
Steve Naroff c77380b46f Incorporate some feedback from Chris...
llvm-svn: 43285
2007-10-24 01:09:48 +00:00
Steve Naroff e670dbb9e6 - Add rewrite rule for @class.
- Add setter/getter to ObjcClassDecl.
- Predefined key runtime functions.

llvm-svn: 43257
2007-10-23 20:20:08 +00:00
Hartmut Kaiser c8107e54d5 Silenced a couple of VC++ warnings.
llvm-svn: 43125
2007-10-18 12:47:01 +00:00
Chris Lattner 8e129c23c8 Move token length calculation out of the diagnostics machinery into
the lexer, where it can be shared.

llvm-svn: 43090
2007-10-17 21:18:47 +00:00
Chris Lattner 02b436a05a Add a new type of lexer: a raw lexer, which does not require a preprocessor
object in order to do its thing.

llvm-svn: 43084
2007-10-17 20:41:00 +00:00
Steve Naroff 0f7a2d2dd1 Predefine all the ObjC goodies from <objc/objc.h>. Removed all the ObjC goodies from the respective test files. Moving forward, it will be very nice to assume these builtin!
llvm-svn: 43077
2007-10-17 17:53:50 +00:00
Anders Carlsson cbfc4b8824 Add support for Pascal strings.
llvm-svn: 42974
2007-10-15 02:50:23 +00:00
Chris Lattner ed2a9eb820 resolve a fixme, by moving __builtin_va_list to a more logical
place and making it correctly parameterized on the target.

llvm-svn: 42830
2007-10-10 17:48:53 +00:00
Chris Lattner 5d45178179 Add a dummy for the id typedef. Steve, plz fill this in. :)
llvm-svn: 42819
2007-10-09 22:58:09 +00:00
Chris Lattner 1f1b0dbc28 Make a significant change to invert the control flow handling
predefined macros.  Previously, these were handled by the driver,
now they are handled by the preprocessor.

Some fallout of this:

1. Instead of preprocessing two buffers (the predefines, then the 
   main source file)  we now start preprocessing the main source 
   file and inject the predefines as a "psuedo #include" from the
   main source file.
2.  allows us to nuke the Lexer::IsMainFile flag and simplify
   Preprocessor::isInPrimaryFile.
3. The driver doesn't have to know about standard #defines, the
   preprocessor knows, which is nice for people wanting to define
   their own drivers.
4. This allows us to put normal tokens in the predefine buffer,
   for example a definition for __builtin_va_list that is 
   target-specific, and a typedef for id in objc.

llvm-svn: 42818
2007-10-09 22:10:18 +00:00
Chris Lattner 98c1f7cfde Switch lexer/pp over to new Token::is/isNot api
llvm-svn: 42799
2007-10-09 18:02:16 +00:00
Chris Lattner ef6b136781 move IdentifierTable.h from liblex to libbasic.
llvm-svn: 42730
2007-10-07 08:58:51 +00:00
Chris Lattner 4894f485c7 implement the Token class in the Lexer.cpp file instead of IdentifierInfo.cpp
llvm-svn: 42728
2007-10-07 08:47:24 +00:00
Chris Lattner c43ddc84a3 improve layering:
Now instead of IdentifierInfo knowing anything about MacroInfo,
only the preprocessor knows.  This makes MacroInfo truly private
to the Lex library (and its direct clients) instead of being 
accessed in the Basic library.

llvm-svn: 42727
2007-10-07 08:44:20 +00:00
Chris Lattner 259716a6e1 change calls to getMacroInfo into hasMacroDefinition() where possible.
llvm-svn: 42726
2007-10-07 08:04:56 +00:00
Chris Lattner d7b971bf3d add a hasMacroDefinition() method to IdentifierInfo, strength reduce a
call to getMacroInfo to call it.

llvm-svn: 42725
2007-10-07 07:57:27 +00:00
Chris Lattner ff067ce555 Remove the PPID bitfield from IdentifierInfo, shrinking it by a word
(because all bitfields now fit in 32 bits).  This shrinks the identifier
table for carbon.h from 1634428 to 1451424 bytes (12%) and has no impact 
on compile time.

llvm-svn: 42723
2007-10-07 07:52:34 +00:00
Chris Lattner a441ca651f First step to fixing a long lived layering violation: this
moves the MacroInfo pointer to a side hash table (which currently 
lives in IdentifierTable.cpp).  This removes a pointer from 
Identifier info, but doesn't shrink it, as it requires a new bit 
be added.  This strange approach with the 'hasmacro' bit is needed
to not lose preprocessor performance.

llvm-svn: 42722
2007-10-07 07:09:52 +00:00
Chris Lattner 5700fab189 simplify the interfaces to create selectors: getSelector can take any
number of arguments now and does the right thing, but the nullary/unary
accessors are preserved as convenience functions.  This allows us to
slightly simplify clients.

llvm-svn: 42716
2007-10-07 02:00:24 +00:00
Chris Lattner f7f34d09e4 simplify some Selector interfaces.
llvm-svn: 42715
2007-10-07 01:33:16 +00:00
Chris Lattner dadc762ac5 Implement DenseMapInfo for Selector, allowing use of DenseMap/DenseSet of
Selector's instead of requiring void* to be used.  I converted one use of
DenseSet<void*> over to use DenseSet<Selector> but the others should change
as well.

llvm-svn: 42645
2007-10-05 20:15:24 +00:00
Steve Naroff e61bfa8bb4 Layering refinements for selectors (suggested by Chris). Specifics...
- Add SelectorTable, which enables us to remove MultiKeywordSelector from the public header.
- Remove FoldingSet from IdentifierInfo.h and Preprocessor.h.
- Remove Parser::ObjcGetUnarySelector and Parser::ObjcGetKeywordSelector, they are subsumed by SelectorTable.
- Add MultiKeywordSelector to IdentifierInfo.cpp.
- Move a bunch of selector related methods from ParseObjC.cpp to IdentifierInfo.cpp.
- Added some comments.

llvm-svn: 42643
2007-10-05 18:42:47 +00:00
Steve Naroff 92866f4fbb Add some comments to MultiKeywordSelector, make all methods private, add a friend, move some methods around.
llvm-svn: 42456
2007-09-28 23:39:26 +00:00
Steve Naroff 8017506d9c Yesterday I discovered that 78% of all selectors in "Cocoa.h" take 0/1 argument.
This motivated implementing a devious clattner inspired solution:-)

This approach uses a small value "Selector" class to point to an IdentifierInfo for the 0/1 case. For multi-keyword selectors, we instantiate a MultiKeywordSelector object (previously known as SelectorInfo). Now, the incremental cost for selectors is only 24,800 for Cocoa.h! This saves 156,592 bytes, or 86%!! The size reduction is also the result of getting rid of the AST slot, which was not strictly necessary (we will associate a selector with it's method using another table...most likely in Sema).

This change was critical to make now, before we have too many clients.

I still need to add some comments to the Selector class...will likely add later today/tomorrow.

llvm-svn: 42452
2007-09-28 22:22:11 +00:00
Steve Naroff 65ca537b55 Fix bug in SelectorInfo::getName() - method buffer needs to be passed by reference.
llvm-svn: 42411
2007-09-27 18:52:21 +00:00
Steve Naroff f73590dbb1 Add SelectorInfo (similar in spirit to IdentifierInfo). The key difference is SelectorInfo is not string-oriented, it is a unique aggregate of IdentifierInfo's (using a folding set). SelectorInfo also has a richer API that simplifies the parser/action interface. 3 noteworthy benefits:
: It is cleaner. I never "liked" storing keyword selectors (i.e. foo:bar:baz) in the IdentifierTable.

: It is more space efficient. Since Cocoa keyword selectors can be quite long, this technique is space saving. For Cocoa.h, pulling the keyword selectors out saves ~180k. The cost of the SelectorInfo data is ~100k. Saves ~80k, or 43%.

: It results in many API simplifications. Here are some highlights:

- Removed 3 actions (ActOnKeywordMessage, ActOnUnaryMessage, & one flavor of ObjcBuildMethodDeclaration that was specific to unary messages).
- Removed 3 funky structs from DeclSpec.h (ObjcKeywordMessage, ObjcKeywordDecl, and ObjcKeywordInfo).
- Removed 2 ivars and 2 constructors from ObjCMessageExpr (fyi, this space savings has not been measured).

I am happy with the way it turned out (though it took a bit more hacking than I expected). Given the central role of selectors in ObjC, making sure this is "right" will pay dividends later.

Thanks to Chris for talking this through with me and suggesting this approach. 

llvm-svn: 42395
2007-09-27 14:38:14 +00:00
Chris Lattner d05e44e74e If we see an invalid #ifdef directive, enter a conditional compilation region
so that we don't emit an error on the #endif.  Suggestion by Neil.

llvm-svn: 42258
2007-09-24 05:14:57 +00:00
Chris Lattner c2d09cfe75 work around bugs and missing features in apfloat.
llvm-svn: 42235
2007-09-22 18:38:30 +00:00
Chris Lattner ec0a6d9be5 Use APFloat for the representation of FP immediates, ask the target
for *which* apfloat to use for a particular type.

llvm-svn: 42234
2007-09-22 18:29:59 +00:00
Steve Naroff 2cd263ff71 Remove SelectorTable/SelectorInfo, simply store all selectors in the central IdentifierTable.
Rationale:

We currently have a separate table to unique ObjC selectors. Since I don't need all the instance data in IdentifierInfo, I thought this would save space (and make more sense conceptually).

It turns out the cost of having duplicate entries for unary selectors (i.e. names without colons) outweighs the cost difference between the IdentifierInfo & SelectorInfo structures. Here is the data:

Two tables:

*** Selector/Identifier Stats:
# Selectors/Identifiers: 51635 
Bytes allocated:         1999824

One table:

*** Identifier Table Stats:
# Identifiers:   49500
Bytes allocated: 1990316

llvm-svn: 42139
2007-09-19 16:18:46 +00:00
Steve Naroff 73d534a2e0 Add support for ObjC keyword selectors.
- Add SelectorInfo/SelectorTable classes, modeled after IdentifierInfo/IdentifierTable.
- Add SelectorTable instance to ASTContext, created lazily through ASTContext::getSelectorInfo().
- Add SelectorInfo slot to ObjcMethodDecl.
- Add helper function to derive a SelectorInfo from ObjcKeywordInfo.

Misc: Got the Decl stats stuff up and running again...it was missing support for ObjC AST's.
llvm-svn: 42023
2007-09-17 14:16:13 +00:00
Chris Lattner 9cf21c5a2c Silence VC++ warnings, patch by Hartmut Kaiser
llvm-svn: 41693
2007-09-04 02:45:27 +00:00
Chris Lattner a263393d11 Add #include required on win32.
llvm-svn: 41686
2007-09-03 18:30:32 +00:00
Chris Lattner 78b6221ff9 Eliminate some VC++ warnings, patch by Hartmut Kaiser!
llvm-svn: 41685
2007-09-03 18:28:41 +00:00
Neil Booth 4a1ee0562d Spaces not tabs.
llvm-svn: 41582
2007-08-29 22:13:52 +00:00
Neil Booth ac582c5ecb Ensure we diagnose long long literals in C90 mode.
llvm-svn: 41581
2007-08-29 22:00:19 +00:00
Chris Lattner ed045421a8 1.0 is double, 1.0F is a float.
llvm-svn: 41412
2007-08-26 03:29:23 +00:00
Chris Lattner f55ab18663 1) refactor some code.
2) Add support for lexing imaginary constants (a GCC extension):

t.c:5:10: warning: imaginary constants are an extension
  A = 1.0iF;
         ^

3) Make the 'invalid suffix' diagnostic pointer more accurate:

t.c:6:10: error: invalid suffix 'qF' on floating constant
  A = 1.0qF;
         ^

instead of:

t.c:6:10: error: invalid suffix 'qF' on floating constant
  A = 1.0qF;
      ^

llvm-svn: 41411
2007-08-26 01:58:14 +00:00
Steve Naroff 7c34817902 Add helper functions Token::isObjCAtKeyword() and Token::getObjCKeywordID().
Convert all clients to the new cleaner, more robust API.

llvm-svn: 41330
2007-08-23 18:16:40 +00:00
Chris Lattner 53e5a38684 Update to match the API from LLVM mainline.
llvm-svn: 41312
2007-08-23 05:22:10 +00:00
Chris Lattner 4c4a245475 Use a smallstring instead of an std::string in FileChanged to avoid some malloc traffic.
This speeds up -E on xalancbmk by 2.4%

llvm-svn: 40461
2007-07-24 06:57:14 +00:00
Chris Lattner f97dbcb4ba Fix a scoping bug that apple gcc doesn't catch for some reason.
llvm-svn: 40450
2007-07-23 22:23:52 +00:00
Chris Lattner 43eafb4ed5 implement a missing feature in the #include handler, where
it did not handle <xyz> headers coming from macro expansions.
This requires special treatment, as the include name is lexed
as multiple tokens, which require reassembly before processing.

llvm-svn: 40418
2007-07-23 04:56:47 +00:00
Chris Lattner 93ab9f134e refactor the interface to Preprocessor::GetIncludeFilenameSpelling,
no functionality changes.

llvm-svn: 40414
2007-07-23 04:15:27 +00:00
Chris Lattner 32e6d64176 fix a bug getting the spelling of an identifier token
that required cleaning.  If the token required cleaning,
don't include the cleaned tokens in the returned length.

llvm-svn: 40410
2007-07-22 22:50:09 +00:00