[cfe-commits] r147664 - in /cfe/trunk/lib: AST/Expr.cpp CodeGen/CGExprAgg.cpp CodeGen/CGExprComplex.cpp CodeGen/CGExprScalar.cpp Driver/Tools.cpp Serialization/ASTReaderDecl.cpp

Eric Christopher echristo at apple.com
Fri Jan 6 16:26:13 PST 2012


On Jan 6, 2012, at 4:21 PM, Chris Lattner wrote:

> 
> On Jan 6, 2012, at 3:08 PM, John McCall wrote:
> 
>> On Jan 6, 2012, at 9:44 AM, Jakub Staszak wrote:
>>> Modified: cfe/trunk/lib/CodeGen/CGExprAgg.cpp
>>> URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/CGExprAgg.cpp?rev=147664&r1=147663&r2=147664&view=diff
>>> ==============================================================================
>>> --- cfe/trunk/lib/CodeGen/CGExprAgg.cpp (original)
>>> +++ cfe/trunk/lib/CodeGen/CGExprAgg.cpp Fri Jan  6 11:44:30 2012
>>> @@ -469,7 +469,7 @@
>>> llvm::BasicBlock *ContBlock = CGF.createBasicBlock("cond.end");
>>> 
>>> // Bind the common expression if necessary.
>>> -  CodeGenFunction::OpaqueValueMapping binding(CGF, E);
>>> +  CodeGenFunction::OpaqueValueMapping(CGF, E);
>>> 
>>> CodeGenFunction::ConditionalEvaluation eval(CGF);
>>> CGF.EmitBranchOnBoolExpr(E->getCond(), LHSBlock, RHSBlock);
>> 
>> The obvious source of breakage here is that this is not at all semantically
>> equivalent.  The original statement declares an object that is destroyed
>> at the end of the scope.  Your replacement creates a temporary object
>> which is immediately destroyed.
> 
> Shouldn't this be reverted then?

It was by Eli earlier IIRC.

-eric



More information about the cfe-commits mailing list