Re: [Bug 375] New - Fix for autoconf >2.5

On Saturday 06 July 2002 02.13, Robert Collins wrote:

> > 1. Instead of modifying the LIBOBJS variable, modify the result
> > cache.
>
> Erk. The problem is changing LIBOBJS at all outside the AC_LIBOBJ
> macro. What we need to do is reproduce that macro for 2.13, without
> colliding with a 2.5x definition.

???

The whole point of this workaround is to not modify LIBOBJS outside of
the standard macros..

> > 2. Then use the existing AC_REPLACE_FUNCS macro to modify
> > LIBOBJS. As the cache is already populated it won't run the test
> > again and simply modifies LIBOBJS if needed as indicated by our
> > custom test.
>
> IIRC LIBOBJS isn't modified. Thats the point of the change. There
> is some reasoning in the changelog/manual. I'm writing this offline
> however, so can't be specific at this time. I'll review and get
> back to the list.

Sounds like a bit of a too big change to be true if AC_REPLACE_FUNCS
no longer modified LIBOBJS directly or indirectly..

In the proposed workaround it does not matter how LIBOBJS is dealt
with within M4, only that these objects gets included in the build
process like expected.

> The handling of object extensions has changed. Automake has begun
> depending on autoconf 2.52 and above. the quoting rules for m4
> macros have been clarified and made consistent. Many macros have
> been replaced with more flexible forms.

> I really don't think it's worth putting time into trying to have a
> 2.5x and 2.13 configure.in when there are what - 6 people? that
> regulary need to create configure, and as the platform where
> configure is created doesn't matter, we can do that anywhere!

I am mostly worried about users who need to patch Squid or otherwise
play around with the code.

Upgrading autoconf or automake on a system should not be taken lightly
as it has a large impact on the ability to build other programs based
on autoconf/automake (quite many GNU or GNU inspired programs relies
a bit too much on being able to use these tools)

Regards
Henrik
Received on Sat Jul 06 2002 - 02:59:03 MDT