llvm-svn: 43188

Fix the previous (short lived:-) FIXME.

I didn't realize that GCC considers this a hard error (I thought it was built-in).

Since it's not, we should simply emit an error.

[dylan:~/llvm/tools/clang] admin% cc -c trivial.m
trivial.m:6: error: cannot find interface declaration for 'NSConstantString'

[administrators-powerbook59:~/llvm/tools/clang] admin% ../../Debug/bin/clang trivial.m 
trivial.m:6:16: error: cannot find interface declaration for 'NSConstantString'
NSString *s = @"123";
               ^
1 diagnostic generated.

llvm-svn: 43157

llvm-svn: 43155

Add a FIXME to an assert.
Change a dyn_cast_or_null back to cast (which is more efficient).

llvm-svn: 43152

llvm-svn: 43151

Tweak a recent fix to UsualArithmeticConversions (made by Chris - r43113). The benefit of this tweak is it guarantees the entire routine operates on unqualified types (which I believe is a bit clearer).  

llvm-svn: 43142

Fix a bug in Sema::CheckConditionalOperands(). When mixing pointers and null pointer constants, we need to promote the null pointer constant (which is an integer) to the pointer type. Test case is self explanatory.

This surfaced yesterday, when compiling test/Sema/cocoa.m on Leopard. Since this has nothing to do with ObjC, it's kind of bizarre this hasn't shown up before. I imagine Cocoa.h on Leopard may have changed recently? 

Thanks to Ted for localizing the bug and giving me a useful AST dump...

llvm-svn: 43114

when comparing "float" and "const float".  This "fixes" the
issue, but may not be the right fix.  Steve, please review.

Testcase here: test/Sema/usual-float.c

llvm-svn: 43113

Fix the following bug...

unsigned char asso_values[] = { 34 };
int legal2() { 
  return asso_values[0]; 
}

The code that creates the new constant array type was operating on the original type.

As a result, the constant type being generated was "unsigned char [1][]" (which is wrong).

The fix is to operate on the element type - in this case, the correct type is "unsigned char [1]"

I added this case to array-init.c, which clearly didn't catch this bogosity...

llvm-svn: 43112

llvm-svn: 43075

Generate code for static variables that don't have initializers. Also, report an error if a static initializer is not constant.

llvm-svn: 43058

llvm-svn: 43051

Add Sema::CheckMessageArgumentTypes()...

llvm-svn: 43050

@selector probably gets this wrong also.

llvm-svn: 43048

Rename SourceRange::Begin()/End() to getBegin()/getEnd() for
consistency with other code.
Start building the rewriter towards handling @encode.

llvm-svn: 43047

llvm-svn: 43046

Remove ObjcMethodDecl::getNumMethodParams/getMethodParamDecl, they aren't used/needed.
Change ObjcMethodDecl::getMethodType to getResultType, to match FunctionDecl.

llvm-svn: 43045

llvm-svn: 43038

Emit diagnostics for methods not found.

llvm-svn: 43037

This fixes a bug Anders noticed.

llvm-svn: 43024

Bad cast...need to use dyn_cast_or_null. Also changed Sema::InitBuiltinVaListType (which had the same bug).

llvm-svn: 43023

Change the type of ObjCStringLiteral from "struct __builtin_CFString *" to "NSConstantString *".

This makes the typecheck much happier. Without this change, the type checker would have to special case "struct __builtin_CFString *". This change does assume the interface for NSConstantString is declared in the translation unit.

I left ASTContext::getCFConstantStringType() around for now (with a comment that says it is currently unused).

llvm-svn: 43021

Move type compatibility predicates from Type to ASTContext. In addition, the predicates are now instance methods (they were previously static class methods on Type).

This allowed me to fix the following hack from this weekend...

// FIXME: Devise a way to do this without using strcmp.
// Would like to say..."return getAsStructureType() == IdStructType;", but
// we don't have a pointer to ASTContext.
bool Type::isObjcIdType() const {
  if (const RecordType *RT = getAsStructureType())
    return !strcmp(RT->getDecl()->getName(), "objc_object");
  return false;
}

...which is now...

bool isObjcIdType(QualType T) const {
  return T->getAsStructureType() == IdStructType;
}

Side notes:

- I had to remove a convenience function from the TypesCompatibleExpr class.

int typesAreCompatible() const {return Type::typesAreCompatible(Type1,Type2);}

Which required a couple clients get a little more verbose...

-    Result = TCE->typesAreCompatible();
+    Result = Ctx.typesAreCompatible(TCE->getArgType1(), TCE->getArgType2());

Overall, I think this change also makes sense for a couple reasons...

1) Since ASTContext vends types, it makes sense for the type compatibility API to be there.
2) This allows the type compatibility predeciates to refer to data not strictly present in the AST (which I have found problematic on several occasions).

llvm-svn: 43009

llvm-svn: 43006

llvm-svn: 43000

Added ASTContext::setObjcIdType/getObjcIdType(), set by Sema.

Also noticed ASTContext::BuiltinVaListType wasn't being initialized to the null type (so I set it).

llvm-svn: 42983

llvm-svn: 42974

- Teach ObjcInterfaceDecl::lookupInstance/ClassMethod to look through protocols.
- Start looking up methods in the global method pools (for "id").
- Start integrating interface types into the type system.

llvm-svn: 42971

Add category lookup (removing a couple FIXME's).
Changed ObjcInterfaceDecl::ListCategories->CategoryList.

llvm-svn: 42968

- Added Sema::AddFactoryMethodToGlobalPool and Sema::AddInstanceMethodToGlobalPool and DenseMaps. This will allow us to efficiently lookup a method from a selector given no type information (for the "id" data type).

- Fixed some funky "}
                    else {" indentation in Sema::ActOnAddMethodsToObjcDecl(). I'd prefer we stay away from this style...it wastes space and isn't any easier to read (from my perspective, at least:-)

- Changed Parser::ParseObjCInterfaceDeclList() to only call Action::ActOnAddMethodsToObjcDecl() when it actually has methods to add (since most interface have methods, this is a very minor cleanup).

llvm-svn: 42957

llvm-svn: 42939

llvm-svn: 42938

llvm-svn: 42936

printf format strings.  Added type checking to see if the matching
width/precision argument was of type 'int'.

Thanks to Anders Carlsson for reporting this missing feature.

llvm-svn: 42933

llvm-svn: 42928

ok class names and streamlined this logic to do the lookup once.

llvm-svn: 42926

Replace one FIXME with another. We handle protocols just fine now. The ObjC decl will only be 0 when we have an error on the ObjC decl. I would prefer we pass in a decl that is marked as invalid. I don't think this is critical to fix now, however I'd like us to be consistent. There are currently many places that don't mark the decl as invalid (which need to be fixed)...

llvm-svn: 42923

llvm-svn: 42917

and hid them.

llvm-svn: 42915

specifiers.

llvm-svn: 42886