[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: AC_FUNC_FNMATCH jamboree patch
From: |
Paul Eggert |
Subject: |
Re: AC_FUNC_FNMATCH jamboree patch |
Date: |
Fri, 7 Jun 2002 00:19:53 -0700 (PDT) |
> From: Akim Demaille <address@hidden>
> Date: 07 Jun 2002 08:58:36 +0200
>
> Paul, BTW, I have always hated AC_FUNC_FNMATCH for not _replacing_ the
> function, as do the others. Do we have to reproduce the error to
> AC_FUNC_FNMATCH_GNU?
No. Sorry, I didn't think of that. How about if we remove the
existing AC_FUNC_FNMATCH_GNU and rename AC_REPLACE_FNMATCH_GNU to
AC_FUNC_FNMATCH_GNU?
> Oh, BTW, I was wondering: the fact that you pass the cache variable
> name is because you want to provide some form of compatibility?
Yes, that was it. Some programs (e.g. a2ps, CVS, RPM) use
ac_cv_func_fnmatch_works to decide whether to replace fnmatch.
Perhaps we should have an upgrade strategy, something like this:
1. Rename the current AC_FUNC_FNMATCH to AC_FUNC_FNMATCH_WORKS, and
define AC_FUNC_FNMATCH to have the same meaning as
AC_FUNC_FNMATCH_WORKS except that it also issues a warning that
its meaning will change eventually and that if you want to avoid
the warning you should use AC_FUNC_FNMATCH_WORKS.
2. In some long-distant future version of Autoconf, modify
AC_FUNC_FNMATCH so that it invokes AC_REPLACE_FNMATCH.
Re: AC_FUNC_FNMATCH jamboree patch, Paul Eggert, 2002/06/06