Re: fix the AC_USE_SYSTEM_EXTENSIONS regression.

[Ralf Wildenhues]

* Eric Blake wrote on Mon, Aug 04, 2008 at 02:59:45PM CEST:
> According to Ralf Wildenhues on 7/31/2008 4:32 PM:
> | Hi Eric,
> |
> | OK to simply revert
> |
> | commit 6add6e92a6136865763edddb9946a8d2469153bb
> | Author: Eric Blake <address@hidden>
> | Date:   Tue Sep 11 17:13:13 2007 +0000
> |
> |     Centralize all system extensions checks.
>
> Unfortunately, now that 2.62 has been released, there are packages that
> are relying on the new combined semantics (pretty much anything
> gnulib-based),

Can you specify this in more detail?  In which way does gnulib, which
aims to only require Autoconf >= 2.59, require the new semantics?

BTW, gnulib has actual users which use it together with 2.59.  It also
has been frequently argued to be much more volatile than Autoconf, which
means pushing an incompatible change in Autoconf which will only impact
gnulib, won't be /that/ much of a problem.  ;-)
Yes, I know I am turning on gnulib's back here a bit.  I don't mean to
force this.

But maybe we can get out of this mess without incompatible changes.
It's not that gnulib requires that AC_AIX incorporate all system
extension checks, right?  Only that AC_USE_SYSTEM_EXTENSIONS incorporate
all checks from AC_AIX (and others), right?

> as well as packages that aren't expecting the combined
> semantics (witness all the bug reports).  I don't think straight reversion
> is an option because it would impact a different set of packages; rather,
> we'll have to find a way to make both camps happy by keeping things
> combined but resolving the barriers that got in the way of older usage
> patterns.

Thanks,
Ralf