Re: Moving on...

Ingo Macherius (Ingo.Macherius@mwe.hvr.scn.de)
Wed, 8 May 1996 20:22:37 +0200 (MDT)


From: Ingo Macherius <Ingo.Macherius@mwe.hvr.scn.de>
Message-Id: <199605082035.WAA13383@ESAMX6.mwe.hvr.scn.de>
Subject: Re: Moving on...
To: bsittler@mailhost.nmt.edu (Benjamin C. W. Sittler)
Date: Wed, 8 May 1996 20:22:37 +0200 (MDT)
Cc: www-talk@w3.org
In-Reply-To: <Pine.SUN.3.91.960508111046.12238A-100000@puce> from "Benjamin C. W. Sittler" at May 8, 96 11:17:51 am

Hi,

> Perhaps a good goal for those of us who would like standard 'serious' web
> publishing would be a network-aware SGML system with freely available
> source. More than one would be even better. Of course, it would need to
> support dsssl or dsssl-o for semantics/rendering, possibly with built-in
> conversion of CSS stylesheets for HTML-like DTDs. 

This is exactly what I'm fiddeling around for some month now. It's kind
of batch driven, entity expanding system using Perl, nsgmls and SGMLSpm
to get work done. Interface will be implemented using CGI.pm and libwww.pm
from the CPAN Mirror. The bad thing that it is copyrighted by
Siemens so I can't give away anything for now. But this is a collaboration
with the Technical University of Clausthal (Germany), so maybe this way will
work. If there is an interest, I may try to convince my marketroid.

> Any volunteers?
Jupp. Please contact me if something envolves.

> (Maybe we could start with one written in tcl/tk using (n)sgmls for the 
> parser)
sgmls is obsolete, nsgmls frees you from stupid low-level things. Believe
me. It's also ported to many platforms, so that's no drawback. I suggest
not to use tcl/tk but Perl, because of the existing Code (Think of 
SGMLSpm and some Earl Hood stuff). Perl is also ported to lots of 
platforms, tcl/tk would bind one closer to Unix.

My problem was handling several dialects of HTML when you re-synthesize
the nsgmls output back to HTML. I do 
$ nsglms foo.sgml | resyth > foo.html
This is done because I benefit from nsgmls entity handling, catalog files
etc.  
When you re-synthesize things you have to be aware of preformatted areas,
attributes IMPLIED by nsgmls, attributes that have URL content, markup that
should have no closing tag (e.g. <LI> or <LINK>), just to name a few.
You definitely need to put this information into 'resynth'. So it may be
a good start to develop a SGML language to summarize dialect-specific
semantics. I'm not firm with DSSSL, so maybe I am not aware that this is
a DSSSL job. But anyway, I don't know any dsssl implementation right now
(beside CoST and several similar aproaches that are not really DSSSL).

Virtually yours,
Ingo 
-- 
Campus:  Ingo.Macherius@tu-clausthal.de      http://www.tu-clausthal.de/~inim
Siemens: Ingo.Macherius@mwe.hvr.scn.de       http://www.scn.de/~inim
 information != knowledge != wisdom != truth != beauty != music == best (FZ)