[XML-SIG] Python / XML / XSLT vs. Cocoon for website server side

Torsten Rueger torsten.rueger at hiit.fi
Fri Aug 15 12:37:09 EDT 2003


Ahh, the ever burning topic of how to build a web application (dynamic 
web site).
Ok I'll chip in. This one on the negative side, the next on the plus 
side:

>> Key decisions:
>>  - All HTML generation done via XSLT
If pain is what you like, that's the way to go.

>>  - Have logic done in XSLT where possible

You do like. Ahh Ouchh.

(Ok I'm joking, all right. You said you never used it and with all the 
books around what you say is a fair choice)

I've only done small-ish (10 pages, 5 posts) application with XSLT 
(Xalan) and XML database (PDOM 
http://www.infonyte.com/en/prod_pdom.html).
The short summary of that is: it's difficult. Unnecessarily so.
Ok, storing xml was easy, but expressing searches in XPath is not 
straightforward. And XSLT isn't easy.

>  My understanding is
> that XSLT is good for recursive pattern matching and generation of
> content from patterns/templates, but not good once you go beyond
> that.

And that's exactly right. So if your xml structure is not exactly what 
your page requires, you've got serious work ahead of you.
Remembering off course that xml does not have a concept of a reference, 
so mapping object dependencies into xml is non trivial.

But the real problem is XSLT is unreadable. Pages you've written only 
days ago seem cryptic.
Maybe more importantly, graphic designers have no chance of 
understanding it.
And it always assumes the dtd of your xml. Change that and you have 
some fun debugging. (So if you do go this way, make it your priority to 
document dtd's)

Anyway, as I said, I'll send another more positive mail on what I have 
found to be beneficial,

Torsten




More information about the XML-SIG mailing list