On Tuesday 27 June 2006 13:18, Akim Demaille wrote:
> >>> "Satya" == Satya <[EMAIL PROTECTED]> writes:
> >
> > Yes, there is a lot of information in the .output file we could use;
> > but this is intended for a human reader;
>
> This file is really obsolete, even for the human. It should be an XML
> file, with XSLT transformations towards HTML and text. I have always
> been annoyed to have to play with C-s under my editor (guess its
> brand) where I want to be able to click on the symbol. There should
> be many many cross-references in that file.
>
> So it would made a lot of sense to design an appropriate XML output
> format, both for other programs and users.
That sounds like a flexible and portable solution to me. I suggest the XML
format is specified in a W3C XML Schema, so it's well defined, is possible to
regression test output, and so on.
However, I would think twice before implementing any XHTML/HTML
export(although XSL-T is an excellent tool for that of course). I would
consider implementing it inside Doxygen because it's a broadly used
documentation framework, supports multiple output formats(PDF, XHTML, etc),
and the parser documentation would be integrated with the rest of the
documentation.
In either case, I think it's clear how useful a modularized approach is, where
whatever interface Bison exposes(XML, API, etc) can be integrated into
different software projects.
Cheers,
Frans
