Re: autoupdate fun

[Stepan Kasal]

Hello,
  I have a patch prepared on my disk.
The problem was that if configure.ac started with
        AC_PREREQ(2.54)
        m4_define([gnumeric_version_epoch], [1])
        ...
then the m4_define disappeared after an autoupdate run.

The fix happen to be quite a big change, but I'm quite sure it makes
things more consistent.  Please see the attached patch.

IIRC, I tested the patch back then, so the only step remaining is to
write an autotest check for this.  I even have an impression that it
existed on my disk at certain moment, but it not there now. :-(

Then we shall test whether the bugs below survive this patch.

Good night for now,
        Stepan Kasal

On Mon, Mar 20, 2006 at 09:33:50PM +0100, Ralf Wildenhues wrote:
> The first bug has been reported against Libtool several moons ago by
> Dalibor Topic.  I've now stumbled over it again and nailed it down to
> AC_LANG_SAVE.  Namely,
>   AC_LANG_SAVE
>   AC_LANG_SAVE
> 
> will lead to m4 looping endlessly.  Smalles useful configure.ac:
>   AC_INIT
>   AC_LANG_SAVE
>   AC_LANG_RESTORE
>   AC_LANG_SAVE
>   AC_LANG_RESTORE
> 
> This prevents users of Libtool to successfully use autoupdate,
> because we use AC_LANG_SAVE twice in libtool.m4.  ATM I don't
> even know whether we can get rid of it (support for older
> Autoconf releases etc).
> 
> 
> The second bug: AS_HELP_STRING arguments are not quoted after
> autoupdate.  I assume this is another bug in AU_ALIAS:
>   AC_HELP_STRING([--enable-foo], [foo bar])
> 
> will turn into
>   AS_HELP_STRING(--enable-foo,foo bar)

autoconf-20050913-au-m4sugar.patch
Description: Text document