xml-dev - Re: Doing it the other way around (Re: transformations)

[Alexey Gokhberg <alexei@bluewin.ch>]

Sean McGrath wrote:

> 
> ... People need to do XML transformation. The look to the W3C
> for guidance. They see XSL/XSLT. Six months later some of these
> people hit a wall - either in terms of functionality, interfacing to the
> outside world, or throughput. This is the way of declarative syntaxes.
> It is why enbedded SQL is more useful than standalone SQL.
> It is why ODBC is the way it is. Perhaps there is a lesson
> here....

Certainly. 

Most of pure XSLT problems can be solved by embedding XSLT into a
powerful scripting engine (e.g., ECMAScript/JavaScript), and by adding
XSLT extensions designed to call scripts from XSLT stylesheets.
Furthermore, scripting engine can provide DOM support to handle
intermediate results and SAX support to filter source/result data and to
integrate various XML transformation components into pipelines. Add
direct support for lower-level components like XPath expressions and
node-sets or XSLT pattern matching, implement regular expressions for
search/match/substitution in text data - and you will get the powerful
transformation engine using only the already existing technologies.

This very approach was implemented in our product - Unicorn XML
Processor (announced yesterday); as I know, several products from other
vendors use the same concept as well.

Regards,

Alexey Gokhberg
Unicorn Enterprises