Re: [xsl] keys and idrefs - XSLT2 request?

Subject: Re: [xsl] keys and idrefs - XSLT2 request?
From: Joerg Pietschmann <joerg.pietschmann@xxxxxx>
Date: Wed, 10 Oct 2001 17:28:33 +0200
"cutlass" <cutlass@xxxxxxxxxxx> wrote:
> this thread reinforces what i mentioned about having one specification
> having an internal dependancy upon another specification

Unfortunately, real world standards have either dependencies to other
standards or significant deficiencies. The real world is complex, you
can't get rid of this easily, except by narrowing your view. A standard
describing a classification of steel rod strengths refers implicitely
(What's a millimeter? degree Kelvin?) and explicitely to other standards
(for example ISO force measurement procedures).

> its a brittle architecture, and all one has to do is add a few versions of
> xslt, xpath, xschema and some other random xml technology like xforms; mix
> them all together and i see a versioning and dependancy nightmare,

I don't know of any successful approach to solve the versioning problem
in face of more than a few interdependant components. It's a real nightmare
for any software development, not only standards. Witness that CORBA
interface versioning is postponed indefinitely, and that's certainly
not because the guys there were dull. Microsoft didn't even attempt
to solve the problem, they just say "upgrade".

This is the way it is, after a while a mess develops, and from the mess
new ideas and a new, beautiful structure emerges, and everyone upgrades.
Look, SGML was fitted to the needs of the authors at the time it
was standardized. However, they didn't get StyleSheets (DSSSL anyone?)
or link standards on time. We built on their experience while
designing XML, XSLT/FO and other standards. There will certainly
emerge a mess again, it's starting already (see, for example, my note
in another post in this thread about XML serializers, or M.Kays
notes about confusing namespace details). Someday we'll have to
start over with something new and more or less incompatible. We
can try to delay this as much as possible by anticipating the
impact of the interfaces between standards on implementing software
and keep them small (high cohesion inside a standards module, low
coupling between them).

> the idea of using XPATH to perform
> validation is cool

Totally uncool. Totally. :-)

Regards
J.Pietschmann

 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list


Current Thread