[help-texinfo] Re: New Texi parser?

[Torsten Bronger]

Hallöchen!

Jean-Louis Leroy <address@hidden> writes:

> Hi Torsten, glad to see you, I had a quick look at your
> texi2latex.
>
> BTW do you know Perl?

Not a single keyword.

>> No.  The conditional parts with @if... are not copied all to the
>> XML file, only some of them.  Comments are missing and two or
>> three other things that I cannot remember at the moment.
>
> My parser throws comments away by default, although it would be
> very easy to subclass it and override parse_c to add Comment nodes
> to the AST. Macros don't show up there either, they are expanded
> via recursive parsing and again it's just a matter of overriding
> the parse_end_macro method.

It's an interesting question which things should be processed and
which things should be just passed to the output file.  For example,
makeinfo's XML backend expands cross references and inserts text
snippets like "See ...".  This is slightly annoying if you process
further the XML file.  On the other hand, some things -- like macros
-- should be expanded (alas, if possible).

>> From today's point of view, the idea to process the Texinfo file
>> with TeX itself was not a very good one.
>
> Whoa, that's how it's done?!

Just for the print output.  Standard Texinfo uses two ways: TeX for
printing and makeinfo for the rest.  Both "programs" share nothing.

>> However, the way via XML is too slow for some purposes, e.g. in
>> Makefiles.
>
> Is speed important? makeinfo --html seems a lot more instantaneous
> than I need. Perl plus multimethods (via Damian Conways'
> Class::Multimethods) surely won't beat C.

Well, not makeinfo is *very* fast.  I tested texi2latex with the
Emacs manual.  My XSLT conversion took minutes, which is
unacceptable in a "make" process.  Dependancies are an issue, too.
XSLT needs an XSLT processor, and Perl is also not as handy as a
small EXE without external dependencies.

Tschö,
Torsten.

-- 
Torsten Bronger, aquisgrana, europa vetus            ICQ 264-296-646