xmlhack: XSLT experts petition against script element

A group of respected XSLT experts have created a petition against the xsl:script element, recently added to the XSLT 1.1 Working Draft.

The element is considered harmful to the portability of XSLT sheets, and contradictory to the statement that "XSLT is not intended as a completely general-purpose XML transformation language".

The petitioners include XSLT implementers (Alexey Gokhberg, Unicorn and Uche Ogbuji, 4XSLT) and experts (Clark C. Evans, Peter Flynn, Francis Norton, Dave Pawson, Tobias Reif, Adam Van Den Hoven) who are very concerned by the addition of a xsl:script element, which they see as far more dangerous for the portability of XSLT sheets than the normalized extension mechanism that was previously included in XSLT 1.1.

Their main concern seems to be that XSLT beginners would jump into this facility to embed Java or Javascript in their sheets without really learning the XSLT subtleties -- a slippery slope down which even confirmed XSLT developers may slide, since XPath functions that cannot be defined using pure XSLT would be definable using Java or Javascript in an xsl:script section.

Among their grievances, they also note that this mechanism would lead the "average user" to "assume that all XSLT processors must support Java and/or Javascript" which is against the principle of language neutrality of XML, dangerous for the authors of XSLT sheets and unbearable for non Java implementations (Unicorn is a C++ implementation and 4XSLT is written in Python).

Even though this petition comes late in the XSLT comments process, Scott Boag and Steve Muench, two members of the W3C XSL Working Group, have encouraged formal feedback and its authors hope to get sufficient user community support before the 15th of March.

The petition can be signed online.

Related stories:

Re: XSLT experts petition against script element (Anatole Tartakovsky - 14:36, 4 Mar 2001)

This proposition makes XSLT unusable for anything beyond academia. XSL evolved as "programmable" alternative to CSS - with "if", "for" statements, templates as "subprograms and functions". Taking away capability to integrate it with other environments diminishes any chance of the language to exist in favor of "algorithmical" languages like XQuery. Do have some respect to your clients, please.

Re: XSLT experts petition against script element (Alexandre Fayolle - 16:49, 2 Mar 2001)

This is exactly what the petition is about: if you want to use XSLT, use it, if you'd rather use roll up your own formalism, do it, using, say, a DOM representation, do it, but outside XSLT.

The problem with people finding it hard to learn XSLT is that they are used to procedural or OO languages (C, Java...), and XSLT is _not_ procedural. I know several people with a strong lisp/prolog background who found it a breeze to learn XSLT. People are reluctant to learn new languages and this is a real shame, because this restricts there ability to think out of the frame.

During the last few months, I've been interviewing candidates who wanted to come and work at Logilab. One of my favourite questions is : "Suppose you're asked to write a program. What language would you use?" Most candidates answer C or Java, without even asking what the program is.

The concern expressed in the petition is that such people would not even take the time to deal with their task in a 'XSLT-ish' way, but jump straight to writing <xsl:script> which is totally WRONG in 95% of cases.

To answer precisely to your devil's advocate's remark, do you *really* think that adding <xsl:script> would help democratize XSLT?

Re: XSLT experts petition against script element (Eric van der Vlist - 18:31, 1 Mar 2001)

How many sites do you think are using XSLT, 0.001%, 0.01%, 0.1%?

(Let's keep playing the devil advocate for a while...)

I would be surprised if this was more than that, and I think we can say that XSLT **IS** still reserved to an elite and you just have to listen on the mailing list to see that people find it hard to start learning XSLT.

That being said, I have signed the petition...

Re: XSLT experts petition against script element (Alexandre Fayolle - 18:16, 1 Mar 2001)

Well, I do not agree. If you just look around, you'll see lots of sites speaking of XSLT, so I think it already has its supporters.

For people who program, it's just another language to learn, and for people who don't, well, obviously they would not use xsl:script either.

Re: XSLT experts petition against script element (Eric van der Vlist - 17:42, 1 Mar 2001)

Yes, this is the idea.

On the other hand, if xsl:script is not implemented, XSLT might stay reserved for an elite and never really take off.

if xsl:script gets implemented xslt will no longer be the best mode for transforming xml documents. we'll be thrown back onto other programmatic methods. probably omnimark or something similar.