[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: New LilyPond website

From: Jean-Charles Malahieude
Subject: Re: New LilyPond website
Date: Fri, 2 Dec 2016 19:50:50 +0100
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:45.0) Gecko/20100101 Thunderbird/45.4.0

Le 02/12/2016 à 02:10, Paul a écrit :
Hi Carl,

On 12/01/2016 12:44 PM, Carl Sorensen wrote:
One thing that is note mentioned in your quote is that texinfo separates
semantics from appearance.  It is this precise separation that allows one
to make big but consistent changes in the appearance of the website with
changes in the CSS.  I'm a firm believer in the principle of separating
semantics from presentation (and we do that with LilyPond, by the way,
which is one of its strengths IMO).

I totally agree about the value of separating semantics and appearance.
Using texinfo does enforce that in a strict way, although well-written
HTML also follows this principle.  HTML for content and CSS for style.
(E.g. one could just write directly the HTML that is generated from
texinfo.)  Maybe I'm missing something or maybe the strict enforcement
is the point?  At any rate, my intention is not to engage in argument or
advocate for anything.

I remember the "traffic" I had to do when the site was in full HTML (back in 2005) and we began with John Mandereau t translate it. Fortunately, I found a macro to use with emacs that "un-htmlized' for the time I typed the French text and "re-htmlized" before pushing any patch. It' hard to read at first sight:

Ces plaques étaient encrées et les reliefs créés par les poinçons et
les découpes retenaient l'encre.

Upgrading to the latest version of texi2any[0] and/or using Haunt would
help, but those are non-trivial endeavors.  The current setup certainly
introduces friction for website work, especially for those who are used
to working directly with HTML.
I believe that we want to avoid working directly with HTML because of its
mixture of semantics and presentation.

I just wish that working with texinfo (for the website) was more
intuitive for contributors who know HTML but not texinfo.  For example,
an HTML element with an id and also a number of classes, all used for
styling it with CSS.  I don't know if you can generate that HTML element
(with both id and classes) from texinfo with our current setup.  If so I
haven't figured it out yet.  Maybe upgrading to the latest texi2any
would help.  Maybe someday I'll have the time to work on it.

I've already given it a try, but get stopped by some errors I don't know how to resolve (I've no knowledge about perl). Three patches are available for anybody willing to help me… I can compile the English version, except that I don't get the TOC sidebar.

Once lilypond-texi2html-init amended, I cannot deal with

Useless use of hash element in void context
Texi2HTML::Config::generate_ly_toc_entries() called too early to check prototype
Prototype mismatch: sub main::normalise_node: none vs ($)
Use of uninitialized value $Texi2HTML::Config::SPLIT in string eq

and tons of warnings, one for each @ref{} encountered, like

out-www/changes.texi:1174: warning: @vindex should only appear at the beginning of a line (possibly involving @rlearning)

comming from the rMANUAL macro definition.

What I do is:
1- uninstall texi2html.1.82

2- create a symlink
  cd ~/bin
  ln -s /usr/bin/texi2any texi2html

In a final stage, occurrences of texi2html might be replaced by texi2any.

3- try building the docs in my local branch dev/texi2any
  make doc LANGS=''

no way until now to build the translations!

4- erase my symlink and reinstall texi2html for usual work on translation branch.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]