[cfe-commits] r64408 - in /cfe/trunk: lib/Headers/limits.h test/Preprocessor/clang_headers.c

[Chris Lattner]

On Feb 12, 2009, at 4:32 PM, Eli Friedman wrote:

> On Thu, Feb 12, 2009 at 3:06 PM, Mike Stump <mrs at apple.com> wrote:
>> +/* The system's limits.h may, in turn, try to #include_next GCC's  
>> limits.h.
>> +   Avert this #include_next madness. */
>> +#if defined __GNUC__ && !defined _GCC_LIMITS_H_
>> +#define _GCC_LIMITS_H_
>> +#endif
>
> Are there any cases where clang doesn't define __GNUC__?  Also, if
> we've gotten here, isn't it guaranteed that _GCC_LIMITS_H isn't
> defined?  Therefore, do we need the #if at all?
>
> Also, probably worth noting an alternative solution I got on IRC: add
> gcc's fixincludes path (something like
> /usr/lib/gcc/i486-linux-gnu/4.3/include-fixed) to
> InitHeaderSearch::AddDefaultSystemIncludePaths.  It looks like gcc's
> solution is more complicated than necessary, though.

Using GCC header files is dangerous.  I don't think it is a problem  
for fix-included header files, but using normal GCC headers implicitly  
makes the produced binary GPL.

-Chris