[cfe-commits] r55635 - in /cfe/trunk: lib/Lex/Preprocessor.cpp test/Parser/block-block-storageclass.c

[Mike Stump]

On Sep 2, 2008, at 9:48 AM, Daniel Dunbar wrote:
> (2) Should we really be defining the attributes to nothing when
> block support is not enabled? This may break working code which
> uses those names.

Yes, it is just part of the design of the feature (programming  
model).  That working code violates the language standard (unless that  
identifier is used in the implementation).  If it is part of the  
implementation, libstdcxx does this (did this) in one or two files,  
those uses will have to be changed.