Re: [PATCH] {maint} distcheck: add support for AM_DISTCHECK_CONFIGURE_FLAGS

[Stefano Lattarini]

Hi Eric.

On Wednesday 15 June 2011, Eric Blake wrote:
> On 06/15/2011 02:18 AM, Ralf Corsepius wrote:
> > On 06/15/2011 09:40 AM, Stefano Lattarini wrote:
> >> On Tuesday 14 June 2011, Ralf Wildenhues wrote:
> >>> * Stefano Lattarini wrote on Fri, Jun 10, 2011 at 05:33:39PM CEST:
> > 
> >>> I would suggest to at least discourage using this in the documentation.
> >>>
> >> ... I agree, and I will make the change soon.  Maybe I can find a
> >> couple of
> >> examples of where the AM_DISTCHECK_CONFIGURE_FLAGS could be useful *and*
> >> legitimate; which would make my change more meaningful.
> > 
> > DISTCHECK_CONFIGURE_FLAGS is useful in situations when a plain
> > "./configure" is not meaningful to a source tree, i.e. when a
> > source-tree mandatorily requires some configuration argument.
> 
> Such a source-tree is violating GNU Coding Standards.  On the other
> hand, automake strives to be useful to more than just GCS-compliant
> packages.
> 
> As for a valid use of AM_DISTCHECK_CONFIGURE_FLAGS, the m4 package
> normally configures --without-changeword, but it is useful to have 'make
> distcheck' exercise the --with-changeword option to ensure that the code
> still compiles for those few people that want to enable the experimental
> configure option.  m4 also uses it to stress-test --program-prefix=g
> (since at one point m4 had a problem where 'make installcheck' was
> assuming it could blindly test "m4" rather than the just-installed "gm4"
> when that renaming option was in effect - see m4.git commit fb04a26fa).
> I've previously been using DISTCHECK_CONFIGURE_FLAGS for that purpose,
> but like the idea of AM_DISTCHECK_CONFIGURE_FLAGS better.
> 
Thank you very much for this real-world example!  Right now I'm in the
middle of trying to write decent documentation fo the new "custom test
divers" feature of Automake parallel-tests, but once I've finished that
I'd like to incorporate your example in my patch.  Or, if you want and
have time, feel free to write a patch on the top of mine, substituting
your real-world example to my artificial.

In any case, it would be nice to refer to the exact m4 version that
exhibits the the behaviour you've described.  So, what is that version?

Thanks,
  Stefano