Re: Merging dev/texi2html [was: Re: Clean history]

[Reinhold Kainhofer]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Am Mittwoch, 20. August 2008 schrieb John Mandereau:
> > That's kind of a deadlock: We don't wan to merge before 1.79 is released
> > on the one hand, while on the other hand, once 1.79 is released and we
> > find problems in texi2html (which happened quite frequently, as the
> > lilypond docs are pushing anything to its absolute limits ;-) ), we are
> > stuck with them...
>
> Texi2html people can do bugfix or pretest releases, can't they?

Sure they can, but I doubt that they would do a quick release of 1.80 if a 
minor bug (but a showstopper for lilypond) is found after 1.79 is released. 

There's one issue I still haven't solved: In the big-page I don't know a way 
yet to assign the correct anchor names, since element_file_name is not 
executed if the file is not split...

> >> * make rules are forbidden in {fr,es,de}/*GNUmakefiles, only variables
> >> specific to each language are allowed there;
> >
> > That was the only way I saw to force creation of the .xref-map files...
> >
> > I had
> > local-WWW-1: $(MASTER_TEXI_FILES) $(PDF_FILES) $(XREF_MAP_FILES)
> > in  make/doclang-targets.make and
> > XREF_MAP_FILES := $(TELY_FILES:%.tely=$(outdir)/%.xref-map)
> > in make/doclang-vars.make, but for whatever reason, the .xref-map files
> > were not generated, so I had to hardcode them in the
> > {fr,es,de}/user/GNUmakefile.
>
> The variable definition with ":=" looks like the culprit.

Okay, I copied this from the lines directly above for the pdf, texi, etc. 
files, where it apparently worked...


> This works after I've hacked the init file a little (my first real
> hack in Perl :-)

Well done ;-) 

> > BTW, looking at Documentation/user/GNUmakefile, I'm wondering why the
> > lilypond-internals.texi rule uses --verbose. This prints about 2 pages of
> > path information during the build...
>
> I usually redirect "make ewb" output to a file, so I don't see the problem.

Sure, but it makes the output of make web a lot harder to read (currently 
it's >>100.000 lines) and scan through to see when something was compiled.

It's no problem, I was just wondering why we are cluttering the make web 
output by things we only need to debug the build system. Similarly, why do we 
call lilypond-book always with --verbose?

In general, we don't need that verbose output, we need it only to debug the 
buildsystem. Can't we have an env variable, say DEBUG_BUILDSYSTEM, and 
use --verbose only in that case (we can then also remove the -q for 
texi2pdf).

Cheers,
Reinhold

- -- 
- ------------------------------------------------------------------
Reinhold Kainhofer, Vienna University of Technology, Austria
email: address@hidden, http://reinhold.kainhofer.com/
 * Financial and Actuarial Mathematics, TU Wien, http://www.fam.tuwien.ac.at/
 * K Desktop Environment, http://www.kde.org, KOrganizer maintainer
 * Chorvereinigung "Jung-Wien", http://www.jung-wien.at/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFIrFdUTqjEwhXvPN0RAm6IAJoCmHNFEM8pILo5xjJ+DMO1UwL74QCfXCOd
d/3J8B1R5MvWPcxRDwV0kq4=
=lcdK
-----END PGP SIGNATURE-----