[PATCH] Change autogen.sh scripts to respect NOCONFIGURE

[Colin Walters]

On Thu, 2012-01-05 at 17:50 +0100, Cyril Brulebois wrote:
> Gaetan Nadon <memsize at videotron.ca> (05/01/2012):
> > > #!/bin/sh
> > > exec xorg-autogen.sh "$@"
> > 
> > Note that I will also have to update configure.ac in each module to
> > mandate the latest version of xorg-macros (as usual):
> > XORG_MACROS_VERSION(1.17)
> 
> So you'll have a check in configure.ac to make sure xorg-macros has
> xorg-autogen.sh, so that autogen.sh can run it? autogen.sh is what
> triggers autoreconf and the look at configure.ac; chicken, egg.

Hmm...but how many people

1) Build an xorg module from git 
2) Are patching a configure.ac or Makefile.am and thus want to rerun
autotools in a package built from tarball

but can't ALSO trivially update xorg-util-macros?  I'm not too familiar
with the workflow of people who hack on X.org related modules, so it's
an honest question.  

In GNOME we recommend jhbuild which makes it trivial to pull in the
latest gnome-common git.  But of course we made a common autogen script
years ago, so that module is pretty widespread; e.g. you can install it
via major distribution packages too.

Regardless we could certainly change the patch to all modules to be:

#!/bin/sh
common_autogen=`which xorg-util-autogen.sh 2>/dev/null`
if test -n "$common_autogen"; then
  exec $common_autogen "$@"
else
  # copy&pasted version here
fi

One question someone watching this discussion might be wondering - why
keep an autogen.sh in each module at all?  There are a few reasons, but
an example one is that it allows a module to individually check for say
a buggy automake version that some of its Makefile.am code triggers, and
bail, but still then exec the common autogen.sh.