Re: killing STDC_0_IN_SYSTEM_HEADERS

[Zack Weinberg <zack at rabi dot columbia dot edu>]

On 06 Jan 2000 19:15:47 -0200, Alexandre Oliva wrote:
>On Jan  6, 2000, Brian Ford <ford@vss.fsi.com> wrote:
>
>> On 6 Jan 2000, Alexandre Oliva wrote:
>
>>> Building gcc on one version of Solaris and using it on another.  Many
>>> people have faced problems after upgrading to Solaris 7, trying to use
>>> GCC built on Solaris 2.6.
>
>> That is funny.  We've been doing just that for seven months or so now
>> without problems.
>
>It seems that my information is just outdated.  gcc 2.7.* would
>present problems for almost any network-related application, because
>lots of headers were `fixinclude'd, and they ended up incompatible
>with those that did not have to be fixed, and were picked from
>/usr/include, regardless of OS version.

Unfortunately, if we were to fixinclude for the "__STDC__ == 0" problem, we
would go right back to touching many headers.  On a 2.6 system:

$ find /usr/include -name '*.h' |xargs egrep -l '__STDC__( - 0)? == 0'
/usr/include/sys/elftypes.h
/usr/include/sys/int_const.h
/usr/include/sys/int_fmtio.h
/usr/include/sys/int_limits.h
/usr/include/sys/int_types.h
/usr/include/sys/signal.h
/usr/include/sys/types.h
/usr/include/ctype.h
/usr/include/dirent.h
/usr/include/grp.h
/usr/include/limits.h
/usr/include/locale.h
/usr/include/pthread.h
/usr/include/pwd.h
/usr/include/setjmp.h
/usr/include/signal.h
/usr/include/stdarg.h
/usr/include/stdio.h
/usr/include/stdlib.h
/usr/include/string.h
/usr/include/time.h
/usr/include/thread.h
/usr/include/unistd.h
/usr/include/varargs.h
/usr/include/widec.h
/usr/include/math.h

It may be that we can live with this - it is not worse than the situation
with gcc 2.7 or on some other platforms.  What do people think of the idea
of allowing the specs to change the definition of __STDC__ ?

zw