xml-dev - Re: XSL Translations using Java Servlets / JSP

[Rick JELLIFFE <ricko@geotempo.com>]

Paul Tchistopolskii wrote:
 
> The <?xml-stylesheet provides some way of binding.
> But this way is not scalable. The scalable way should
> allow:
> 
> rendering any document with any stylesheet without
> changing the document itself an any point of processing.
> 
> ( validating any document with any DTD without changing
> the document itself  - but that's not as important as it is
> with the stylesheets ).

On the other hand, scalability is a universal criterion for ajudging
merits of a technology, in the same way that "simplicity" and "power"
are not virtues except in particular contexts.   Is a RAID disk array
too simple? Is it too powerful (overkill)?  Without some context
information those questions are unanswerable.

If I want to publish one particular type of document with one particular
style, then the xml-stylesheet PI is fine.  

Isn't this just our old question of when embedded markup is appropriate
(e.g., see Ted Nelson's article in the XML Journal put out by OReilly
editid by Dan Connolly, for one instance of this argument, which goes
further than it needed to to make its point) rather than out-of-line and
external markup?

Our question re-appears with linking, with stylesheets, with
multilingual documents, with schemas; the patterns may be thoguht of as
"inline markup" versus "out-of-line markup". But there is more: there
are scoping levels each of which can represent a decision point for
"inline" versus "out-of-line".  Viewing at the document-level scope, a
spreadsheet PI represents use of "inline markup" (i.e. within the same
document) rather than "out-of-line markup" (i.e. in some other
document). However, viewing at the element-level scope, that spreadsheet
PI is "out-of-line" rather than "inline".  

In XML we can distinguish four nested boundaries where the
"inline/in-band" versus "out-of-line/out-of-band" scoping can be seen:
  * element scopes
  * entity scopes
  * document scopes
  * global (i.e., everything else)
though there is clearly some need to split the last into some
package/bundle versus global system (I guess relative URLs are being
used to do this at the moment).  (And SGML would have some level of
GROVE above document, I guess: is the re-serialization of a GROVE a
package?)

SGML 86 was very heavy on allowing document scoping: on the other hand,
it did allow elements to directly contain documents (SUBDOC) to allow
nested scoping (e.g. of identifiers, document type, etc). HTML tried the
opposite extreme and pretty much only had element scoping.  Which one is
simpler? which one is more powerful?  You cannot tell without context;
without seeing the particular problems being solved and the particular
skills and technologies available.



Rick Jelliffe

***************************************************************************
This is xml-dev, the mailing list for XML developers.
To unsubscribe, mailto:majordomo@xml.org&BODY=unsubscribe%20xml-dev
List archives are available at http://xml.org/archives/xml-dev/
***************************************************************************