Re: [bug-gnulib] Re: <stdint.h> on IRIX-5.3 & SGI-cc

[Martin Neitzel]

> > #  if  > 32 ||                                      <=== flagged error here
> This is valid ANSI C and ISO C 99 code. When you don't pass "-w2" to your
> compiler, does it show an error or a warning about this?

Omitting -w2 doesn't change anything.  (Not surprisingly, since -w2 doesn't
soften the effect of true errors but only the effect of warnings.)

I agree with you that a strictly conforming compiler would have to close
its eyes on the murky "> 32 ||" in the context of the preceeding "#if 0".

None of the IDO cc's option to switch to other preprocessor variants
("-acpp", "-oldcpp") will improve the situation.  I will invest some
time to resolve this specific issue cleanly in some way.


> >         cc -DHAVE_CONFIG_H -I. -I. -I..  -I../intl  -Ino/include  -w2 -g -c 
> > md5.c
> > cfe: Error: /usr/include/sys/types.h, line 211: redeclaration of 'int8_t'; 
> > previous declaration at line 78 in file './stdint.h'
> >  typedef        signed char     int8_t;
> 
> Should be fixed since Friday.

No, I still get this error.

The responsible stdint_.h code is:

#if address@hidden@                     -==> #if !0    in stdint.h
typedef signed char    int8_t;
#endif

I'll try to have a look at the details.  I'd would help me to know
what gnulib's approach is or, more important perhaps, how gnulib
is used for the CVS project.  If somebody could check the following
list, that might get me oriented more quickly:

[ ]  (a) CVS uses gnulib to built upon the C89 standard on any platform.
[ ]  (b) CVS uses gnulib to built upon the C99 standard ...
[ ]  (c) CVS uses gnulib to built upon the first POSIX standard.
[ ]  (d) CVS uses gnulib to built upon the latest POSIX/SUSv3 standard.
[ ]  (e) CVS ...

                                                                Martin