[cfe-commits] r150318 - in /cfe/trunk: include/clang/AST/ include/clang/Sema/ lib/AST/ lib/CodeGen/ lib/Parse/ lib/Sema/ lib/Serialization/ test/CXX/temp/temp.decls/temp.variadic/ test/Sema/ test/SemaTemplate/

[Jeffrey Yasskin]

On Tue, Feb 14, 2012 at 12:44 PM, Richard Smith <richard at metafoo.co.uk> wrote:
> On Sat, Feb 11, 2012 at 3:51 PM, Sebastian Redl
> <sebastian.redl at getdesigned.at> wrote:
>>
>> Author: cornedbee
>> Date: Sat Feb 11 17:51:47 2012
>> New Revision: 150318
>>
>> URL: http://llvm.org/viewvc/llvm-project?rev=150318&view=rev
>> Log:
>> Represent C++ direct initializers as ParenListExprs before semantic
>> analysis
>> instead of having a special-purpose function.
>>
>> - ActOnCXXDirectInitializer, which was mostly duplication of
>>  AddInitializerToDecl (leading e.g. to PR10620, which Eli fixed a few days
>>  ago), is dropped completely.
>> - MultiInitializer, which was an ugly hack I added, is dropped again.
>> - We now have the infrastructure in place to distinguish between
>>  int x = {1};
>>  int x({1});
>>  int x{1};
>
>
> g++4.6 has a bug where it silently accepts "t({...})" as if it were a
> brace-or-init-list initializer in a constructor's member initializer list.
> By the law of nature that for every g++ "extension" there is an equal and
> opposite libstdc++ bug, libstdc++ 4.6's <bitset> uses this syntax to
> initialize its member array in C++11 mode. Is it feasible to downgrade the
> error in this particular case to an ExtWarn?

Please also file gcc bugs in cases like this. They're generally happy
to try to fix gcc-trunk so more instances don't pop up in the future.