Re: XSLT for Mashups

Michael Kay wrote:


>> What would it offer over and above doc() and
xml-over-http ?

> I think the challenge is that simple RESTful services can
>
indeed be done using doc(), and anything more complex gets you
> into issues
of side-effects.  But Florent Georges and others
> are working on ideas, and
I'll be more than happy to support
> them if they produce something good.
Thanks, Mike.  But clearly, REST cannot be supported by doc().
You can issue
simple GET requests, but REST is based on more
HTTP verbs than GET.  Not
mentioning headers yet.

  And once you have the technical tools to issue
general HTTP
requests, you can go further and provide libraries that wrap such
requests within a more simple API (dedicated to REST WS, SOAP WS,
Google APIs,
etc., taking care of authentication based either on
standard or proprietary
headers, OAuth, OpenID, etc.)

  Aside being able to write web services
clients in XSLT (and
other XML technologies like XQuery and XProc), there is
also the
interesting possibility of writing web services themselves using
those technologies.  And again the principle is the same:
providing the
technical tools to write a stylesheet to respond to
HTTP request (so the other
way around: on the server), then
building up on those tools more specialized
frameworks (XRX, REST
WS, part of the WS-* stack, etc.)

  And yes,
side-effects are a key point.  I think XSLT has all
the machinery to handle
them (like the output state) and XQuery
will have too (with XQuery SX).  But
the output state is maybe
too much restrictive (e.g. regarding function
calls); for now,
just using the HTTP Client naively has showed up to not cause
any
problem.

  Rendez-vous in XML Prague for more information ;-)

  Regards,
-- 
Florent Georges
http://www.fgeorges.org/