Re: The XSL-List Digest V2 #191

Subject: Re: The XSL-List Digest V2 #191
From: Paul Prescod <paul@xxxxxxxxxxx>
Date: Sun, 27 Jun 1999 12:52:00 -0400
Scott Ferguson wrote:
> With each draft, 'XQL' needs more XSL declarations to work.  'XQL' now depends on
> XSL to work for ids, variables, extension functions, keys, locale, namespaces, and
> the doc() function.  

What you're saying is that every version allows a richer binding between
the embedding environment and the query. This is no different than SQL or
OQL which also allow the defintion of (e.g.) extension functions.

> The namespaces and extension functions are the real killer.  

How would you propose to avoid those? One could, theoretically define an
alternate syntax for the query language that allows the namespaces to be
spelled out as URIs, but surely you don't think that that should be

And extension functions are crucial for making the query language
extensible -- whether in XSLT or anywhere else. SQL and OQL are both
designed to allow the environment to add extension functions also.

> My
> main concern, really, is that the next draft will include more dependencies as XSLT,

XSL has been declared feature-complete. It has also been declared that
most of XSL's query language will be unbundled for use in XPointer. 

> solely to avoid the dreaded xsl:script and xsl:eval tags, becomes a full-fledged
> functional language.

You haven't established any relationship between the environment-embedding
features of XSLT and XSLT's lack of support for inline scripting. It looks
like a red herring to me.

> To make XQL work with XSL, I've got an environment class, Env, that you need to fill
> with variables, key declarations, id declarations, locale, and pass it as an
> optional argument.  Each XSL declaration adds some more random fields to the Env
> class.

I don't see this as a big problem. In SQL and OQL these extensions are
just embedded in the database. Probably future XML databases will also
offer this option.

> Is this a major problem for XSLT?  Probably not.  My API getting grungier and more
> random with each new XSLT draft is well down on the list of important issues :-).
> But I believe Microsoft has a similar API, and the DOM level n + 1 may eventually
> support XQL.  So my concern isn't entirely selfish.

I don't see your environment object as being grungy. It falls out fairly
naturally from the description of "context" in the XSLT draft. XSLT has
one syntax for setting up the context and of course your library will have
a different one. 

 Paul Prescod  - ISOGEN Consulting Engineer speaking for only himself

Perhaps the war in Kosovo would get more press if it were directed by
George Lucas.

 XSL-List info and archive:

Current Thread