Re: autoconf 2.5, 2.13, and Vim

[Akim Demaille]

>>>>> "Paul" == Paul Eggert <address@hidden> writes:

>> From: Akim Demaille <address@hidden> Date: 12 Jun 2001 16:44:09
>> +0200
>> 
>> Including if we just use the $ac_includes_default?

Paul> I'm not sure I understand this question.  Presumably
Paul> ac_includes_default should work on all hosts, even AmigaOS or
Paul> other non-POSIX hosts, and this means ac_includes_default should
Paul> wrap sys/types.h inside '#if HAVE_SYS_TYPES_H'.  If all tests
Paul> include ac_includes_default, and if ac_includes_default includes
Paul> a wrapped sys/types.h, then there's no reason for a test to
Paul> include sys/types.h explicitly.  Hence all tests will wrap
Paul> sys/types.h.  This is not a maintenance burden, as the wrap
Paul> occurs only in one place.

Hence my proposal!

Paul> However, if we have a conscious policy to wrap all include
Paul> directives, then many changes are needed (more than the changes
Paul> that you proposed, as we'll need to wrap all the other headers,
Paul> too), and I'm not sure it's worth the hassle.  

Anyway that's a step in the direction I want to go, that of
autosystem.  I really want to address this issue in the future of
Autoconf, and it seems to me that the right direction is to have
configure use system.h for its tests.  Which means, unifying.

Paul> These changes will ripple through to user-visible recommended
Paul> behavior.  For example, we'll need to change the documentation
Paul> for AC_HEADER_SYS_WAIT, change the example for AC_CHECK_MEMBER,
Paul> etc.

The real question is: are we introducing incompatibilities. I don't
think so, but I do agree I did not study deep enough.  Nonetheless, I
don't think so.

Paul> Personally I'd like to hear other people's opinions.  So far we
Paul> seem to have one "wrap everything" (you), 

That's somewhat unfair :)  My position is: converge, and hence, since
for some we want to wrap, wrap'hem all.

Paul> one "don't bother with any of this" (Russ), and one "wrap just
Paul> the tests that should succeed on non-POSIX hosts" (me).  So far
Paul> I seem to be holding the center, but perhaps that's just because
Paul> we haven't heard the way-out opinions.  :-)