Re: Small document and sample HTML and DocBook

[Patrice Dumas]

On Tue, Jan 06, 2009 at 01:25:38PM +0000, Noah Slater wrote:
> 
> I think we should avoid XHTML because it causes more trouble than it's worth. 
> If
> this is a controversial opinion on this list I am happy to spend some time
> digging out citations for you. (but I'd rather not, obviously)

texi2html is more controversy-proof than makeinfo since it is easy 
to customize the output. In the past I did an init file for texi2html
(which would require updating), that turned the output into valid
transitional xhtml. But I don't think that anybody ever asked, I think
that I did that on my own. The aim was not to have a html compatible
xhtml, but to have valid xhtml on its own.

> If it's not too hard, I would aim for full HTML strict output. If this is an
> insurmountable task, anyone who wants strict can use HTMLTidy. Whatever the
> case, I would choose one and stick with it.
> 
> > It would need much more css.
> 
> What would need more CSS?

strict HTML doesn't have <u> or <font>, not width in table... 

I attach the result of the use of http://validator.w3.org/
with a strict dtd, and the original file produced by texi2html.
With Transitional, it is valid.

> I have been working on a default stylesheet for my GNU manual that balances
> æsthetics with readability and I would love to contribute this to texi2html 
> and
> work with you on getting it broadly useful for distribution.

I am not that much interested in the stylesheets themselves, but
rather in having appropriate class attributes in the output to let
anybody do its own stylesheet. That being said shipping .css files
to show examples of use would be nice, even more if the output is 
nice ;-)

> My current draft is available at:
> 
>   http://periplum.org/dev/gnu/publish/trunk/doc/publish.html

Indeed it looks good.

> This is currently produced with:
> 
>   makeinfo --enable-encoding --html --no-split --css-ref=style.css

I could use it to check that texi2html output is compatible with
makeinfo --html output css-wise.
 
> > Also I would like to rework the css support in texi2html, to be more
> > compatible with what makeinfo currently do, and to enhance (and fix) css, 
> > but
> > after the merge. Currently info output is way higher on my list (and
> > progressing rather neatly).
> 
> Not sure what this means. How can I help out?

Myabe this is not the answer you are looking for but, as you already 
know, the idea is to replace C makeinfo with texi2html. For that texi2html
has to converge to makeinfo to be strictly compatible. This, in turn, 
implies:

* generating info with texi2html. This is my priority right now, and 
  after some delay, I am devoting all my free software time slots to 
  that task, and it is progressing well.

* having texi2html converge toward makeinfo regarding the command 
  behaviour. This should not be a big deal:
    + command line switches and processing are almost the same, only 
      little change is needed
    + error messages and lines in error messages formatting have to be
      synced
    + minor other issues

* convergence in html output. 
    + As a first step texi2html output should take what is clearly 
      superior in makeinfo --html output. 
    + Also remove unneeded inconsistencies, for example in CSS classes.
    + and last decide the default html output when invoked as 
      makeinfo --html, including missing css classes, progress toward
      strict html... such that the html output is at least as good as
      makeinfo --html output

* using a gettext-like handling of internationalization

--
Pat