[Top][All Lists]

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

Re: [Andrew Suffield <address@hidden>] Bug#157887: autoconf: various imp

From: Paul Eggert
Subject: Re: [Andrew Suffield <address@hidden>] Bug#157887: autoconf: various imperfectly formed functions in c.m4
Date: Tue, 24 Sep 2002 16:42:10 -0700 (PDT)

> Date: Tue, 24 Sep 2002 19:03:27 -0400
> From: Harlan Stenn <address@hidden>

> I'm working on one application where the package Requires gcc, and the
> design spec Requires "-Wall -Werror".

Weird -- the design spec requires you to be slaves to the GCC police?

But anyway, surely it the spec requires this only of your program, not
of your "configure" script's contents.  So you should be able to put
-Wall -Werror in AM_CFLAGS or something like that.

> However, with 2.54 I am now seeing message about how I should't be using
> CFLAGS (or a whole bunch of others), but should be using the AM_ versions
> instead.

As far as I know, Autoconf doesn't know or worry about AM_CFLAGS.
That's Automake's job.  Perhaps you've run into an interoperability
problem (wouldn't be the first time), but I'd like to know details
about how to reproduce it.

> So what's the preferred way to write software that uses automake and
> autoconf given that some of these variables need to be amended during the
> configure process?

The preferred way is to not amend them during the configure process,
but to amend them (if needed) during the build.  It has never been
that portable to assign to CFLAGS at random points in your configure
script.  (I know, I know, I've done it myself -- but I've usually
regretted it afterwards.0

reply via email to

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