RFC: May <inttypes.h> be used by testcases of the C-testsuite?

[Joseph S. Myers]

On Thu, 5 May 2005, you wrote:

> Would it be ok to make use of inttypes.h in testcases of the C-testsuite? 

No, it is a C99 hosted-only header which GCC does not provide and many 
targets do not provide.  The exception is gcc.dg/intmax_t-1.c which 
deliberately uses the system inttypes.h in the hope of detecting whether 
GCC's intmax_t is wrong.

Feel free to work out how to fix bug 448 (implementing <stdint.h> in GCC) 
and then use <stdint.h>.  Geoff Keating had a problem implementing his 
pragma approach, that the pragma would need to define macros but would be 
handled too late to do so with lexing up front as used by the C++ front 
end <http://gcc.gnu.org/ml/gcc/2004-11/msg00277.html>.  I think you want 
to implement Zack's proposal to split GCC's internal include directories 
<http://gcc.gnu.org/ml/gcc/2004-11/msg00254.html> first to reduce problems 
with compatibility with existing system headers (though you still need to 
do a fair amount of work to keep the definitions the same as those systems 
already have and to ensure compatibility with <inttypes.h> on systems 
which only have that).

-- 
Joseph S. Myers               http://www.srcf.ucam.org/~jsm28/gcc/
    jsm@polyomino.org.uk (personal mail)
    joseph@codesourcery.com (CodeSourcery mail)
    jsm28@gcc.gnu.org (Bugzilla assignments and CCs)