Re: [xsl] XSLT Dead?

Subject: Re: [xsl] XSLT Dead?
From: "Robert Koberg" <rob@xxxxxxxxxx>
Date: Mon, 16 Apr 2007 19:49:40 -0400
Dang... just reading this now. I would love to comment, but Abel said what I would have said.
:)
One more thing though, I wonder if some folk do a disservice to people like Karl. It definitely seems there are people who know that almost everything can be done in XSL, but don't mention that perhaps it shouldn't.


I mean a comment like, "if it is XML, it should be transformed with XSL..." It just seems wrong.

-Rob


On Mon, 16 Apr 2007 18:59:23 -0400, Abel Braaksma <abel.online@xxxxxxxxx> wrote:


Karl Stubsjoen wrote:
>So, is XSLT dead?

What do you mean by "dead"?  That no developer will use it or that no
supplier will make XSLT available?  Or something else?

Dead, as in there are better technologies to explore, like off the shelf solutions, a custom calendar control for example.

Interesting choice of example. I work daily with XML, XSLT and many other XML related technologies, but if I need a calendar control, I'd investigate my host system and choose what best suits my needs. That way you'd have a prebuild DHTML / JavaScript calendar control for some intranet, a prebuild .NET or ActiveX control for a desktop system and a prebuild paper version for on the wall.



I, the XSL enthusiast I am, would think WAY different than a .NET solution available for purchase. If tasked with "give the user a calendar so he/she can pick a date from", well knowing what I know about XML/XSL, I'd opt to write my own and pre-gen the data for the calendar as XML, probably using C# to dynamically write the XML out for the transformation. But you see, my solution would almost always include XML and an XSL transformation which is opposite of the developer looking for a solution already written for .NET.

Interesting, again. I've seen a programmer choosing Assembler for writing a Windows GUI app. He needed 4K where I needed 2M. The difference was only that his took 3 months and mine only 2 days. Let alone the maintenance issues. Do not use XSLT when there are other tools that can do the job better. I can write a lexical analyzer for C++ in XSLT, but really, I won't.



I would like ALL opinions. I myself, have made the concious choice to avoid the MS .NET gui path. I have made the choice to use XML and XSL where ever I can.

I can understand the rather critical attitude of your co-workers. "where ever I can" is a dangerous and often costly choice.



They are looking for the precanned solution. Its purchase vs. build, and they are opting to purchase.

I agree with them: why re-invent something if it is already there? In all but a few situations, precanned solutions will be the most valuable and cost-effective ways, even if you have to leave out some of the very nice options you would've build yourself in your own tool. It is a common misunderstanding with programmers that they think they can build something better or more useful where other companies have spend tens of thousands of man hours before you.




How does a "drag and drop GUI designer" relate to a result-tree
construction process based on multiple XML inputs in a declarative
construction-rule stylesheet?

They don't. It is a shift in thinking. Much like the calendar example above, I make the concious choice to think results (solutions) in XML and XSL.

I like XSLT, I really do. But I'd never choose XSLT where an imperative language would make my life easier. Yes, XSLT is "Turing complete" and I can probably make the new tamagotchi generation with it. But "can make" is not a synonym for "should make".


Really, choose the best language for the job. Choose XSLT if you have to go from one structured input (XML or CSV or SAP ISU or whatever) to another structured format (HTML, XSL-FO, CSV etc). Don't use it for the things it isn't meant for, you can use your time for building other creative and ground-breaking solutions.

Good luck with your colleagues,

Cheers,
-- Abel Braaksma



-- Using Opera's revolutionary e-mail client: http://www.opera.com/mail/

Current Thread