RE: About Constructions rules

Subject: RE: About Constructions rules
From: "Didier PH Martin" <martind@xxxxxxxxxxxxx>
Date: Mon, 19 Jul 1999 22:06:24 -0400
Hi Brandon,

Brandon said:
   I'm not sure I understand what you're trying for here.  I guess I
always sort of "knew" it, but I have to thank Dr. Steve Newcomb for
explicitly pointing out to me that most SGML processing models (DSSSL
included) are totally "data-driven".  The document that serves as
input to the process totally drives the processing.  So, I have to
ask, why would you want to create a construct that, somehow, does
something all on its own, without being somehow "triggered" by the
source document?  What can you see being able to do with something
like this that can't be done with the approach shown above?

Didier says:
Yes you are right, DSSSL is based on an event based mechanism and the
document drive what's happening. The grove navigation, in fact, provides the
events or nodes to be pattern matched by rules. A push model is more related
to templates. In that case, the script contains the output document as a
template and then some part of the template is modified by the script
constructs. The more I think about his, the more I come to think that a
<style-specification-body> should be restricted to event driven construction
rules and that a other sgml element could be set for templates like for
example <template-body> itself a child of <template>. It would be more clean
that way and two kind of approach could be used to produce an output:
a) style based and constructing flow objects
b) template based and incorporating information from the source document or
doing computation to insert computed elements (like queries, etc..). Anyway,
this is a snapshot of my actual "in progress" thinking.


Brandon said:
   Having done the "go in and fix or extend someone else's code"
thing, I know just how you feel. :)  However, knowing that you guys
are getting into the guts of {Open}Jade is a great thing, because it
gives me confidence that the community will have an opportunity to put
into OpenJade all the things we've been wishing Jade had all along.
   On that note, I'm starting to think *very* seriously about making
an attempt at implementing the transformation language in Jade.  I'm
just dying to play with it, and it seems like Jade would be the best
place to get it working quickly, given its stability and widespread
use.
   I've barely begun to lay out a plan of attack for this, and may not
do too much more until I hear from Henry Thompson, who has said that
he'll get me some information on his test implementation based on his
DSC toolkit, as that will probably give me a pretty good head start.
   I can hack my way through C++ reasonably well, but my time is also
quite limited, so if there are some reasonably skilled C++ programmers
out there who would be interested in contributing to this effort, let
me know, as I'd be more than happy to coordinate the efforts of other
people, in addition to my own, to pull this off.
   So, OpenJade architects, what do you think of trying this?  Can you
see a place for it in OpenJade?

Didier says:
I would say simply: welcome aboard Brandon. This would be a pleasure to work
with you. I think implementing the transformation part is what is missing to
get a complete product. We'll post the result of our archeological research,
so this could help better understand the openJade internals.

regards
Didier PH Martin
mailto:martind@xxxxxxxxxxxxx
http://www.netfolder.com


 DSSSList info and archive:  http://www.mulberrytech.com/dsssl/dssslist


Current Thread