[tex4ht] question on split levels for tex4ht

[Michal Hoftich]

Hi Nasser,

> 1 - split at \part
> 2 - split at \chapter
> 3 - split at \section
> 4 - split at \subsection
> 5 - split at \subsubsection
> 6 - split at \paragraph
> For document classes without \chapter (like article):
>
> 1 - split at \part
> 2 - split at \section
> 3 - split at \subsection
> 5 - split at \subsubsection
> 6 - split at \paragraph
> -----------------------------------------
>
> My question is: Why does in article, it skips split level 4, and
> goes from 3 to 5? I find this confusing. May be there is reason
> internally when tex4ht does this?
>

I am not sure why this works in this way, I guess it would be more
logical for classes without \chapter to skipt the option "2". But I am
not sure about changing this, as it would be a quite a big breaking
change. I didn't notice this because if I break pages at all, then I
usually break pages only at the top level, so either at \section or
\chapter.


> I'd like to suggest for some future version that this
> option should be redone. May be add new option calling it
> "split_at" so current method stays backward compatible.
>
> The new option will have the form
>
>         split_at="name"
>
> where name can be "part" or "chapter" or "section" or "subsection"
> and so on.
>
> So one can now do
>
>      make4ht   foo.tex 'html,mathjax,split_at="section"'
>
> If string does not work for value, it can just name, as in
>
>      make4ht   foo.tex 'html,mathjax,split_at=section'
>
> Not only is this much more clear than
>
>      make4ht   foo.tex 'html,mathjax,3'
>
> But with the new option, one do not have to worry changing the command
> to get the numbers right, if they change from article to
> book.

I am not sure about this. If you take a look at the html4.4ht file,
you will find that the code for the numerical options is quite
complex. For example, search for \:CheckOption{7} and then try to find
\CheckOption{1}. It is quite a lot of code. And it is duplicated for
every document class. It would be another several hundreds of
kilobytes if we added that.

Best regards,
Michal