XML, XSL, XSLT, XPath support?

[Frans Englich]

On Thursday 07 April 2005 13:15, Matt Rogers wrote:
> On Thursday 07 April 2005 08:48 am, Frans Englich wrote:
> > On Wednesday 06 April 2005 16:53, Andy Beltran wrote:
> >
> > Hello Andy,
> >
> > > Is there any plans for Konqueror to support XML, XSL, XSLT, and XPath?
> >
> > The one who did it planned it. A hand from one who has knowledge about
> > the language is welcome.
> >
> > You're invited to help out with developing XPath/XSLT. One of the first
> > steps is to convert the XPath stack in kdenonbeta/xpath to KDE's
> > "next-generation" DOM implementation, kdom, found in kdenonbeta/kdom. How
> > to fetch these sources, is described here:
> > http://developer.kde.org/source/anoncvs.html
> >
> > The KDOM developers have a small readme file with notes etc about
> > development, which I have attached.
> >
> > Questions can be posted on this list, or privately. Feel also free to
> > drop in on #ksvg on the open project IRC network(KDOM developers).
> >
> >
> > That's the plans for XSLT support.
>
> Bring yourself back to reality for a minute. You propose to make
> kdenonbeta/xpath work with KDOM when KDOM isn't even used in KHTML yet, and
> nobody has said it will be. Maybe you should think about finishing
> kdenonbeta/xpath and make it work with current KHTML first, as I'm sure
> that would be much more useful for both Andy and anybody else who wants
> XML, XSL, XSLT, and XPATH support in KHTML now rather than later.
> Otherwise, you're asking somebody to invest time in something that might be
> cool but isn't guaranteed to be used, and it sounds like Andy could really
> use support for XPATH in KHTML now rather than later.

Then we have different views of what is reality, and hence disagree.

For example:

I personally consider XSLT a large project. My estimation is that if someone 
started on XSLT now as his/her main project on a volunteer basis, it would be 
ready for khtml use far after a khtml port to kdom finishes.

I neither see it as unsure whether khtml will use kdom. From an XML 
perspective the only sensible thing is to base on a rich DOM implementation. 
For example, if someone would start on XSLT, the wisest thing would probably 
be to complete DOM 3 Core and DOM 3 Load and Save(and duplicate efforts? To 
semi-port KDOM?). Similarly, XPath is a helper technology(mostly), and what 
it is to help needs a proper DOM -- that's KDOM.

If one is a bit strategic, one prepares and lies the groundwork for XSLT 2.0. 
That's about ten W3C specifications. libxslt's implementation is about 50 000 
lines of code.

In other words, from my perspective I'm being economical with the sparse man 
power we have. I don't think you can write something usable -- even if it 
means one have to do a port to kdom later on -- and get it shipped with KDE 
3.5. I wouldn't do it for another reason: doing a sub-optimal result for the 
sole reason of bringing a fast result.

But it's nevertheless a question of estimations and will/man power, so there's 
no absolute answers.


Cheers,

		Frans