Re: [help-texinfo] New Texi parser?

[Karl Berry]

    However, I went for the two things that seemed a bit more complex,
    namely macros and file inclusion.

Yes, macros are the worst complication when it comes to parsing.

@value can also be troublesome.

The @defun commands are a pain because they have bare braces, unlike
most everything else.

Then there is all the stuff which is significant in some contexts but
not others, such as commas and newlines.

    How large is the community? 

Based on the mailing list subscriptions and traffic, not that large as
far as as developers go, but not zero, either :).

    Karl find a little time to point out the tricky parts? 

I can try.

    It's tempting to start an open-source project to have a kit of
    flexible texi tools. 

Yeah.  Torsten Bronger recently wrote a texi2latex program based on the
XML output from makeinfo (as I recall).  That seemed to work out fairly
well.  It might be the best way to write additional "flexible" tools.

Here is my basic problem: unless any new version is a replacement for
makeinfo, it turns into another program that parses Texinfo source that
has to be maintained separately.  That's not good.  And something isn't
a replacement for makeinfo unless it is really 100% compatible -- there
are too many existing manuals which depend on finicky details.  All
these stodgy-yet-necessary requirements have pretty much stymied me ...

Thanks,
Karl