Steve Naroff [Wed, 24 Dec 2008 20:59:21 +0000 (20:59 +0000)]
Add explicit "fuzzy" parse support for Microsoft declspec.
Remove previous __declspec macro that would effectively erase the construct prior to parsing.
Douglas Gregor [Wed, 24 Dec 2008 02:52:09 +0000 (02:52 +0000)]
Keep track of template arguments when we parse them. Right now, we don't actually do anything with the template arguments, but they'll be used to create template declarations
Douglas Gregor [Wed, 24 Dec 2008 00:01:03 +0000 (00:01 +0000)]
Correct the order in which we cope with end-of-class-definition
semantics and improve our handling of default arguments. Specifically,
we follow this order:
- As soon as the see the '}' in the class definition, the class is
complete and we add any implicit declarations (default constructor,
copy constructor, etc.) to the class.
- If there are any default function arguments, parse them
- If there were any inline member function definitions, parse them
As part of this change, we now keep track of the the fact that we've
seen unparsed default function arguments within the AST. See the new
ParmVarDecl::hasUnparsedDefaultArg member. This allows us to properly
cope with calls inside default function arguments to other functions
where we're making use of the default arguments.
Made some C++ error messages regarding failed initializations more
specific.
Douglas Gregor [Tue, 23 Dec 2008 21:31:30 +0000 (21:31 +0000)]
When determining whether a class type has a const copy constructor, be
sure to look at all of the results returned by name lookup. Fixes
<rdar://problem/6465262>
Douglas Gregor [Tue, 23 Dec 2008 21:05:05 +0000 (21:05 +0000)]
Don't push OverloadedFunctionDecls onto the chain of declarations
attached to an identifier. Instead, all overloaded functions will be
pushed into scope, and we'll synthesize an OverloadedFunctionDecl on
the fly when we need it.
Ted Kremenek [Tue, 23 Dec 2008 18:41:34 +0000 (18:41 +0000)]
PTH: Use 3 bytes instead of 4 bytes to encode the persistent ID for a token.
- This reduces the PTH size for Cocoa.h by 7%.
- The increases PTH -Eonly speed for Cocoa.h by 0.8%.
Ted Kremenek [Tue, 23 Dec 2008 02:52:12 +0000 (02:52 +0000)]
PTH:
- Encode the token length with 2 bytes instead of 4.
- This reduces the size of the .pth file for Cocoa.h by 12%.
- This speeds up PTH time (-Eonly) on Cocoa.h by 1.6%.
Ted Kremenek [Tue, 23 Dec 2008 02:30:15 +0000 (02:30 +0000)]
PTH:
- In PTHLexer::Lex read all of the token data from PTH file before
constructing the token. The idea is to enhance locality.
- Do not use Read8/Read32 in PTHLexer::Lex. Inline these operations manually.
- Change PTHManager::ReadIdentifierInfo() to PTHManager::GetIdentifierInfo().
They are functionally the same except that PTHLexer::Lex() reads the
persistent id.
These changes result in a 3.3% speedup for PTH on Cocoa.h (-Eonly).
Ted Kremenek [Tue, 23 Dec 2008 01:30:52 +0000 (01:30 +0000)]
PTH:
- Embed 'eom' tokens in PTH file.
- Use embedded 'eom' tokens to not lazily generate them in the PTHLexer.
This means that PTHLexer can always advance to the next token after
reading a token (instead of buffering tokens using a copy).
- Moved logic of 'ReadToken' into Lex. GetToken & ReadToken no longer exist.
- These changes result in a 3.3% speedup (-Eonly) on Cocoa.h.
- The code is a little gross. Many cleanups are possible and should be done.
Douglas Gregor [Tue, 23 Dec 2008 00:26:44 +0000 (00:26 +0000)]
Don't explicitly represent OverloadedFunctionDecls within
DeclContext. Instead, just keep the list of currently-active
declarations and only build the OverloadedFunctionDecl when we
absolutely need it.
This is a half-step toward eliminating the need to explicitly build
OverloadedFunctionDecls that store sets of overloaded
functions. This was suggested by Argiris a while back, and it's a good
thing for several reasons: first, it eliminates the messy logic that
currently tries to keep the OverloadedFunctionDecl in sync with the
declarations that are being added. Second, it will (eventually)
eliminate the need to allocate memory for overload sets, which could
help performance. Finally, it helps set us up for when name lookup can
return multiple (possibly ambiguous) results, as can happen with
lookup of class members in C++.
Next steps: make the IdentifierResolver store overloads as separate
entries in its list rather than replacing them with an
OverloadedFunctionDecl now, then see how far we can go toward
eliminating OverloadedFunctionDecl entirely.
Douglas Gregor [Mon, 22 Dec 2008 05:46:06 +0000 (05:46 +0000)]
Add support for calls to overloaded member functions. Things to note:
- Overloading has to cope with having both static and non-static
member functions in the overload set.
- The call may or may not have an implicit object argument,
depending on the syntax (x.f() vs. f()) and the context (static
vs. non-static member function).
- We now generate MemberExprs for implicit member access expression.
- We now cope with mutable whenever we're building MemberExprs.
Eli Friedman [Sun, 21 Dec 2008 01:15:32 +0000 (01:15 +0000)]
Fix for PR3246: an empty clobber list is the empty string, not a null
string.
That said, we should probably try and track down the correct clobber
lists for the targets that don't have them (PPC, ARM, and Sparc),
so that we can generate correct code.
Douglas Gregor [Sat, 20 Dec 2008 23:49:58 +0000 (23:49 +0000)]
Add support for member references (E1.E2, E1->E2) with C++ semantics,
which can refer to static data members, enumerators, and member
functions as well as to non-static data members.
Implement correct lvalue computation for member references in C++.
Compute the result type of non-static data members of reference type properly.
Eli Friedman [Sat, 20 Dec 2008 23:11:59 +0000 (23:11 +0000)]
Make VLAs usable, and make basic usage work correctly. Also, add a
simple test that actually does VLA codegen.
Note that despite the fact that the alloca isn't in the entry block, it
should dominate all uses; this is guaranteed by the restrictions on goto
into VLA scope in C99.
Strangely enough, name of ObjC class is not encoded into the
full encoding of the class which has an ivar of pointer to this
class. Its name is encoded in the type for the ivar in the
ivar-list metadata. This patch conforms to the above rule.
Zhongxing Xu [Sat, 20 Dec 2008 06:32:12 +0000 (06:32 +0000)]
Lazy bingding for region-store manager.
* Now Bind() methods take and return GRState* because binding could
also alter GDM.
* No variables are initialized except those declared with initial
values.
* failed C test cases are due to bugs in RemoveDeadBindings(),
which removes constraints that is still alive. This will be fixed in later
patch.
* default value of array and struct regions will be implemented in later patch.
projects / clang / log