Re: more translation breakage

[Graham Percival]

On Wed, Jan 13, 2010 at 01:19:30AM +0100, Francisco Vila wrote:
> 2010/1/8 John Mandereau <address@hidden>:
> > Le vendredi 08 janvier 2010 à 12:15 +0000, Graham Percival a écrit :
> >> grr.  If you add new files in master, check that you add them properly.
> >>
> >> file from VC not distributed: 
> >> lilypond-2.13.11/Documentation/es/search-box.ihtml
> >> file from VC not distributed: 
> >> lilypond-2.13.11/Documentation/hu/search-box.ihtml
> >
> > Fixed.  I have no easy mean to check this, sorry.
> 
> I've just pushed the translated Essay into the lilypond/translation
> branch but I don't know how to make new files distributable or what
> GNUMakefile to copy into essay/.

Given that John's fix *should* have worked (adding stuff to
one of the stepmake/ or make/ files), but didn't, I have to say
that *nobody* really knows how to make new files distributable.
I'm *fairly* certain that .texi and .itexi files are distributed
by default... no wait, that's wrong, we needed
EXTRA_DIST = web.texi

*shrug*

This will give you no joy, but compared to all the other breakage
and poorly-explained stuff in the build system, I'd rank this as
priority-medium at most, with no fix (from me) expected for at
least 2 months.


John: the way to test this is to do "make dist", then untar
out/lilypond-2.13.11.tar.bz2.  If there's a file in git that isn't
in that tarball, there's a problem.  I'm sure there's a handy way
of listing all files in zip and comparing it to ls -R of the
uncompressed tarball (actually, you probably don't even need to
uncompress it), but you're on your own for figuring out those
options.

As soon as 2.12.{3,4}-{1,2} is out, I want to get 2.13.11 out.  If
you merge stuff to master before then, please go to the trouble of
doing the above test.  If you wait until after 2.13.11 for the
merge, I don't care if you don't test it -- I'll test it with the
GUB stuff, and add the missing files myself.  As long as this
doesn't cause *unexpected* problems in GUB, at a time that isn't
convenient, I think the most practical thing is just to let GUB
encounter the error after an hour of release-building.

Cheers,
- Graham