Re: stdint vs cycle-check.h

[Mark D. Baushke]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Paul Eggert <address@hidden> writes:

> "Mark D. Baushke" <address@hidden> writes:
> 
> > is it desirable to make those modules depend on stdint.m4 and avoid the
> > HAVE_STDINT_H macro?
> 
> Yes, I think so, in the long run.  But in the short run, stdint.m4 is
> not yet reliable enough (as you're finding out with SGI), and some
> packages don't want to rely on it for that reason.
> 
> For cycle-check.h I'd think this wouldn't be a problem, since
> cycle-check.h uses this Autoconf-recommended sequence:
> 
> # if HAVE_INTTYPES_H
> #  include <inttypes.h>
> # endif
> # if HAVE_STDINT_H
> #  include <stdint.h>
> # endif
> 
> and this should work on SGI.

With lots and lots of warnings, yes. 

The HAVE_STDINT_H is defined, even though it is useless and there is an
stdint.h which is generated and includes the useless SGI version and
generates warnings for everything because of the include of <inttypes.h>
which duplicates everything being put into the generated stdint.h file.

> If this doesn't actually work for you, or if there is a problem with
> some other gnulib source file that uses HAVE_STDINT_H, please let us
> know.

In CVS, the windows-NT does not supply a <stdint.h>, so we use stdint.m4
to create one. However, the lib/*.c files do not build because there is
not one available when we tell the truth about HAVE_STDINT_H being
undef, yet if we tell the truth, then the generated stdint.h is wrong
because it uses something like this:

#if @HAVE_STDINT_H@
#  include @FULL_PATH_STDINT_H@
#endif

and there is actually no full path to <stdint.h> to be used.

        -- Mark
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (FreeBSD)

iD8DBQFEmGM4Cg7APGsDnFERAvQlAJ9v/v0x36SN1qEMkBBA7x07g9Z66wCfZzry
Sy29t8z28pLWYY2jkbxydps=
=gH5T
-----END PGP SIGNATURE-----