RE: [docstandards-interop-discuss] A concern. Semantic mismatch and how to address it.

["David RR Webber \(XML\)" <david@drrw.info>]

Dave,

Your approach below is TO BE AVOIDED AT ALL COSTS!  Ocean boiling alert!!

We know this fails to provide agile exchanges - and entails DOUBLE TRANSFORMATIONS - in and out.

You should look at OASIS CAM work for how we are approaching this for transactional XML.  This allows you to state the rules for the pieces you care about - and flex on the rest of it. XPath provides loose linkage to structure parts without being overly restrictive on where what occurs.

And CAM syntax reuses XPath and xslt so it does not invent much of anything itself.  It also allows implementers to add their own custom extensions for local needs.

I like what Michael was saying yesterday - DITA + xhtml + extensions to handle obvious document items that xhtml missing (things like ToC markup) should be enough for a very workable V1.0 - applying the 80:20 rule - and staying well away from ocean boiling.

DW

"The way to be is to do" - Confucius (551-472 B.C.)


> -------- Original Message --------
> Subject: [docstandards-interop-discuss] A concern. Semantic mismatch
> and how to address it.
> From: "Dave Pawson" <dave.pawson@gmail.com>
> Date: Wed, April 11, 2007 5:03 am
> To: docstandards-interop-discuss@lists.oasis-open.org
> 
> In terms of scope mainly.
> 
> If (big if) my assumptions on goals are approximately right.
> And assuming some form of hub language / Esperanto into which and from
> which all transforms occur.
> And assuming XML semantically marked up inputs and outputs.
> 
> 
> Concern.
> 
> How to avoid developing a markup language which is the Nirvana of
> SGML, i.e. addresses
> all needs in all scenarios.
> 
> It's been tried and failed so many times.
> 
> go as big as TEI.
> go as specialised as docbook
> go as re-usable as Dita.
> 
> The use case scenario would be some geek attempting to map 'my markup'
> to hub markup
> and always coming up with 'nearly matches' but not quite, what do I
> do. I can see this happening even with a point to point? E.g. TEI to
> docbook nearly matches etc.
> 
> The issue can be avoided by making the hub language the 'master', i.e.
> the sender
> determines what, from the hub language is the best match and defines
> the map.
> Similarly the output transform specifies the hub to target map,
> knowing that it
> won't be a perfect fit.
> 
> Would this provide real benefit?
> 
> 
> regards
> 
> 
> 
> -- 
> Dave Pawson
> XSLT XSL-FO FAQ.
> http://www.dpawson.co.uk
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail:
> docstandards-interop-discuss-unsubscribe@lists.oasis-open.org
> For additional commands, e-mail:
> docstandards-interop-discuss-help@lists.oasis-open.org