Re:New Project Proposal (or not..)

["Daniel M. German" <dmg csg uwaterloo ca>]

Stephen Rust twisted the bytes to say:

 Stephen> On Thu, 25 Mar 1999, Daniel M. German wrote:
 >> Stephen,
 >> 
 >> SGML is a huge language. It might be better if you deal with XML
 >> instead of SGML. SGML is inherently difficult to parse and to support
 >> in its totality. XML is simple to parse, its DTDs are slim. 

 Stephen> I guess my main goal of the project was to have something similar to a
 Stephen> 'netscape composer' type of feel, where you throw some text on a screen,
 Stephen> format it around the page like you want, with colors, borders, tables,
 Stephen> type of things, and then be able to export it to a lot of different
 Stephen> formats like the sgmltools provide capability of.  I agree, supporting the

IMHO, you should not go this way. XML is structurally oriented, not
presentation oriented. I am assuming you know about DTDs. In a DTD you
specify elements and restrictions on where this elements appear
(notice that a "Tag" is not an element, instead it is a specification
of where the element starts and ends). An element does not have any
typesetting attached to it. That is given by a stylesheet. 

While it is not a problem to go from XML to typeset text (such as a
WYSIWIG screen) it is very difficult to go the other way around. 
Even in plain HTMl is it difficult to go from "typeseted" text to HTML
elements and that's why Word processors and Netscape suck and their
HTML is usually crap but it looks good on the screen. :) They use HTML
more as a presentational langauge than a structural one.

Instead of provind colors and tables and such, a XML editor should
provide a list of potential elements available at a given point
within the document --according to a given DTD. The typesetting engine
should then present the elements in whatever way  they are supposed to
be. For example, an element XXX can be translated to an flow element
"Table"  and an element YYY that can only appears inside XXX can be
translated to flow element "TR" . But maybe somebody else can edit the
document as if "XXX"  was "H1"  and YYY "h2". That's the beauty of
XML.

The other thing to consider is what to do with XML documents that are
not valid with respect to the DTD (valid XML doc). XML provides a
"escape": documents that look like good XML but have no DTD attached
(well-formed documents). 

Users sometimes do not want to be force to have a valid document at a
certain stage and a good editor should jump from validating mode to
non-validating mode easily.

 Stephen> whole SGML language would be quite a task.  And I've read that XML is sort
 Stephen> of like a re-birth of SGML, with possibility for some widespread
 Stephen> acceptance, so maybe that is the way to go.  

XML is the way to go (IMHO). You don't want even to learn the
complexity and uglyness of SGML. That's why SGML editors are soo few
an expensive. XML has a lot of potential.

 Stephen> At least to start with I don't want to create a hugely complicated
 Stephen> process.  But the possibility to create some simple formatted text is
 Stephen> highly appealing, with the help of the GUI.

It will be very good to create a simple editor that can take a DTD and
just allow insertion of valid XML (even without typesetting). It
should provide feedback of the structure of the document (sort of a
tree view of the document) and allow the user to see what is available
at a vertain point in the document). If you use emacs, you should
check the pgsml mode of it. If you can do that for XML we will be very
grateful. The typesetting engine can then be added at another stage
and maybe, using CORBA, done by the gnome WordProcessor.

 Stephen> I like the idea of being able to show the code, and the result at the same
 Stephen> time.  I do have more reading to do regarding typesetting, stylesheets,
 Stephen> etc. however.  But for example in 'MS Access' where you have a window with
 Stephen> your bare SQL code, and then another window with your choices and layout
 Stephen> which created the SQL code, sounds like an interesting way to show it.

Yeap. I recommend you read the XML spec at the w3 consortium
(www.w3.org). You can find also info on XSL, the stylesheet language.

 Stephen> - How large a project would this be?
 >> 
 >> It depends on what your goals are. If you are shooting for SGML (as
 >> defined in the stardard) it might take quite a loooong time.

 Stephen> Definitely a subset of something (sgml or xml) to start.  At least in the
 Stephen> short run, and the motivation for the project, I'm not looking for full
 Stephen> implementation, just the ability to create some simple documents quickly.

Look into the specs and check for "Well formed documents". If you can
come out with a well-formed documents editor, it is a very good
start. Then the next stage will be to add XML validation capabilities.

 Stephen> Thanks for the feedback,
 Stephen> Steve

--
Daniel M. German                  "If Microsoft ever does applications
   Linus Torvalds ->               for Linux it means I've won."
http://csgwww.uwaterloo.ca/~dmg/home.html
dmg@csg.uwaterloo.ca