> you're being too polite; they haven't progressed at all.  they would
> have been done months ago, but when i tried to follow how the pages
> are generated by the lilypond tarball, i finally despaired after
> following the endless maze of makefiles calling other makefiles ad
> infinitum, by which time the real world (tm) had caught up with me and
> demanded my attention again.  about a week ago, i tried it again with
> the same result.
> 
Ouch, that's not good.  You shouldn't have such troubles, but when
you do, don't hesitate to send a mail and include the screendump
(actually, i'm expecting to get one of you shortly!)
We should fix the problem, or add documentation; it would be a
shame to have more people run into these problems, and maybe 
be put off by them.

Anyway, it should be not too hard to build the website and whatever
approach we'll take, i guess it would be good if you could build the
website as it is now yourself.  I'll try to explain the process in
brief.

The website consists of three parts: Documentation, Examples and
Doc++ of the sources.  The whole webpage should be built by the
toplevel command:

     make website

However, when this doesn't work, you can try to build the individual
parst separately:
 
     make htmldoc      # the Documentation part

This will recurse into the directory Documentation.  Basically, all
documentation there is (.yo, .tex, .doc, .bib, ...) is converted to
html or to ps (yodl2html, tex/latex && dvips, bib2html).  All html
and .ps is generated in directories out-www.  Each subdirectory
of Documentation can be htmlized by 

      make CONFIGSUFFIX=www local-WWW

For some directories (i guess Documentation and 
Documentation/topdocs) there is a hand-made index.yo, 
for others (tex/ and man/) the out-www/index.html is generated
by a python script (buildscripts/mutopia-index.py) that lists
all stuff in out-www.
The index.yo in the topdocs directory gets copied to toplevel,
and becomes the lilypond homepage.
If all goes well, you should find a htmldoc.tar.gz in the toplevel
out-www directory.

The examples part is done in a similar way:

     make examples    # the Examples part

recurses through the directories input and mutopia.  Each
GNUmakefile has a variable (EXAMPLES=...) that lists what
examples to build.  LilyPond is run on the listed examples,
and .txt .png, .ps and .midi output is 'produced' in out-www.
Again, mudela-index generates an out-www/index.html from
what's there.

> try to pull the resulting documentation into my own setup, or to try
> to contribute to the tarball directly so that the result of running
> "make website" is my fancier page.  if we were to go through with
> 
I guess there are three approaches:

    1 handmake a whole new website, copy and incorporate parts
      of lily's website as it is 'now'.
    2 make a static frontend that is in effect a redo of all
'index.html'
       files that are now in the website, so that examples and doco
       are always up to date
    3 make a new target 'fancywebsite'

Somehow i've got the feeling that we should be doing either 1 or 3.
Option 2 seems to be a recipy for desaster, missing links etc.
I would hope we could do 3; obsolete webpages are almost as
bad as webpages full of broken links.

> registering lilypond.org as has been suggested in the past, what would
> 
The biggest problem is getting another nameserver, apparently.  As
soon as we have one, our friend tiggr can get us the domain.  I'd taken
care of that with digicash about two months ago, but alas...

> be better for it, a more static page that may be slightly behind the
> times in some respects but more dependable, or one that is generated
> from the lilypond source and thus up-to-date but perhaps broken in
> some places?
> 
Perhaps the best option would be a combination of the above 2 and 3;
i.e., build the fancy webpage from the lilypond sources, but do that
only every now and then (e.g, at most once a month).  It should include
all important information that shouldn't be broken, faq, mailing lists, 
urls, and perhaps a nice interface to all examples and mutopia. As an
extra, there would be some links to 'experimental stuff' that may be
broken once in a while, showing (some of the) latest examples.

Jan.

Reply via email to