bug-gnulib
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Bug-gnulib] argp.m4 missing


From: Simon Josefsson
Subject: Re: [Bug-gnulib] argp.m4 missing
Date: Mon, 18 Aug 2003 21:42:56 +0200
User-agent: Gnus/5.1003 (Gnus v5.10.3) Emacs/21.3.50 (gnu/linux)

Bruno Haible <address@hidden> writes:

> Simon Josefsson wrote:
>
>> The argp stuff is not finished yet (latest version of patch sent to
>> libc-alpha), so the following may need to be modified.  Still, it is
>> what I'm using, and it works on the 15 or so systems my buildbot use.
>>
>> The argp modules must be modified too, to add some more prerequisites,
>> and to change the AC_FUNC_ARGP name into gl_ARGP as previous
>> discussed.
>
> Thanks, I've committed your patch.
>
> But I'm wondering why you do AC_CHECK_HEADERS([argp.h]) since the
> module provides "argp.h" by itself, and HAVE_ARGP_H is nowhere used?

Hm.  The intention was to check in the system argp.h if that version
is acceptable, and if so disable the argp package.  But perhaps we
should rely only on the ELIDE_CODE a'la getopt stuff only.  The
problem with doing that is that the gnulib argp code will be compiled
for machines that run glibc 2.1, 2.2, 2.3 (which doesn't contain a
_GNU_ARGP_INTERFACE_VERSION declaration), which is a waste.

The m4 could do (pseudo code):

if HAVE_ARGP_H && HAVE_ARGP_PARSE_FUNC then
   #define _GNU_ARGP_INTERFACE_VERSION 1

so the gnulib argp code would be elided, since the gnulib argp will
(at least initially) be API compatible with the already deployed argp.

I hope you understand my thoughts, I realize this isn't very clear.
(Which, in part, is because I haven't implemented the ideas yet.)





reply via email to

[Prev in Thread] Current Thread [Next in Thread]