Re: [xsl] Re: Spam:[xsl] RE: String manipulation in XSLT

[Wendell Piez <wapiez@xxxxxxxxxxxxxxxx>]

At 01:26 PM 10/20/2005, Jay wrote:
> XSLT 1 doesn't have much support for string manipulation, because XSLT 1
> is really about manipulating XML and no one perceived much need for string
> manipulation. XSLT 2 has far greater string manipulation ability because
> it became apparent that those features were useful to a lot of people.

I wasn't (quite :-) present at the creation either, but I do think 
Jay rather overstates the point: it's not really true that "no one 
perceived much need for string manipulation". Rather, it was 
envisioned that there'd be other tools (Perl, anyone?) better suited 
to that. XSLT 1.0 was supposed to be a clean, lean and mean subset 
and optimization of *some* of the things that made its predecessor, 
DSSSL, so powerful, while fixing those things about DSSSL (the 
syntax) that made it impossible for most people.

Jay is correct that requirements for upconversion (the kind of 
transformation for which string manipulation is generally essential) 
were explicitly left out of XSLT 1.0 (look at the Rec). Its 
assumption was that your XML was already clean and well-designed, and 
all you were calling XSLT to do was downconvert it, probably for 
presentation. It was never supposed that that is all you'd ever want 
to do with your XML, or that there aren't also problems in how you 
get the good XML to begin with. These were (are) smart people who 
assumed there'd be more than one tool in your box.

Then too, XSLT 1.0 was deployed with mechanisms to allow vendor 
extension functions to fill these gaps, and provide a pathway forward 
where 1.0's minimalism clearly fell short.

The designers of XSLT 1.0, I'm sure, have been surprised by many 
things. But the fact that people need to do upconversions, but XSLT 
1.0 doesn't do them well, isn't one of them. One surprise has perhaps 
been the degree to which the market has succumbed to the Golden 
Hammer effect ("I need to code a routine to square a circle: let's 
use XSLT!") -- and it's probably true that the expansion of XSLT's 
scope in 2.0 has a lot to do with that.

Jeni Tennison wrote a whole book about using XSLT 1.0 for Things For 
Which It Wasn't Intended. "XSLT On the Edge". I recommend it -- both 
to see what can be done, and to see what perhaps shouldn't be.

Cheers,
Wendell