Re: Problems with compiling the documentation

[John Mandereau]

Le samedi 11 août 2007 à 13:14 +0300, Till Rettig a écrit :
> Hello everybody,
> 
> I just tried to compile the documentation from git master branch. First 
> it gave an error about a missing epsf.tex file that collated-files.tely 
> in input/manual/out-www needed. I found that my version of texlive had 
> only installed xepsf.tex. So I linked it to the name epsf.tex and 
> everything was fine. Would it be possible to let search for both 
> versions of the file? It seems xepsf.tex if compatible.

Which Texlive version do you have?  I have Texlive 2007 on my box, and
it includes epsf.tex.  Maybe you need to upgrade to 2007 or rerun
Texlive setup to install the package providing epsf.tex.


> Then compilation stopped at buildscripts/post-www.py whith following 
> message:
> 
> /usr/bin/python ./buildscripts/www_post.py LilyPond 2.11.29 
> ./buildscripts ./out-www "offline"
> Traceback (most recent call last):
>   File "./buildscripts/www_post.py", line 14, in <module>
>     package_name, package_version, buildscript_dir, localedir, outdir, 
> targets = sys.argv[1:]
> ValueError: need more than 5 values to unpack

Argument localedir is obviously missing in www_post invocation, which is
quite strange.  GNUmakefile.in from Git HEAD contains

$(PYTHON) $(buildscript-dir)/www_post.py $(PACKAGE_NAME) $(TOPLEVEL_VERSION) 
$(buildscript-dir) $(top-build-dir)/Documentation/po/$(outdir) $(outdir) 
"$(WEB_TARGETS)"

Please check you have the same command line in GNUmakefile.  If it's OK,
then there might be a problem with $(top-build-dir); in the case you
have moved the source and build tree, make sure you have rerun
configure.

Please report any succesful or unsuccesful experience, so you can start
updating the docs ;-)


> I saw the change making the amount of values six was introduced in 
> February, and I have successfully compiled with that, so is it about my 
> python? I have 2.5 installed (with Ubuntu 7.04) but also 2.4 available. 
> Should it explicitly call 2.4 and that would make the mistake disappear? 

I use Fedora 7, which comes with Python 2.5, and build the docs almost
every day, so Python version is certainly not a problem.

Cheers,
John