Re: Documentation translation: first patch

[Jan Nieuwenhuizen]

Han-Wen Nienhuys <address@hidden> writes:

>> rsynced from web-root, and not untarred from the docball?
>
> no, it's rsynced from uploads/webdoc/v<VERSION>/
>
> this has to be done, because output distance computations need
> information from previous versions.

Hmm, then I misread lilypond.org.py.  To bad, I checked-in something
that produces a plain doc ball, and an extension-stripped
target/<native>/build/out/web-root into GUB.

>> The extension stripping is a bit tricky to move out
>> of add-html-footer, as I have documented there.
>
> I missed that, I think; which commit is this?

No idea.  I also only see commits on master on my rss feed?

It's just this snippet from add-html-footer.py that is needed for
content negotiation.

    # Strip .html, .png suffix for auto language selection (content
    # negotiation).  The menu must keep the full extension, so do
    # this before adding the menu.
    if content_negotiation:
        page = re.sub 
('''(href|src)=[\'"]([^/][.]*[^.:\'"]*)(.html|.png)(#[^"\']*|)[\'"]''',
                       '\\1="\\2\\4"', page)

    # Add menu after stripping: must not have autoselection for language menu.

Problem is that if this is moved out of add-html-footer.py, we'd have
to do something ugly or clever (probably both) to avoid stripping the
extension in the language menu (the menu will not direct to another
language page if the file name url is not fully specified).

Clever/ugly things could be: do re.sub only on part until some footer
mark, or add some sort of do-not-strip-me attribute to an url...

We'd still have to build the .html's twice.

Jan.

-- 
Jan Nieuwenhuizen <address@hidden> | GNU LilyPond - The music typesetter
http://www.xs4all.nl/~jantien       | http://www.lilypond.org