Some thoughts from the peanut gallery.

> - Should we try to reuse Saxon's processor? It is even possible? Is 
licensing compatible?

I'm curious what this even means?  Saxon is a a competitor of Xalan's. Are 
you proposing to wrap Saxon with Xalan API's?

> - Should we try to reuse Eclipse's Psychopath XSLT/XPath 2.0 processor? 
It is even possible? Is licensing compatible?

As far as I can tell Psychopath implements XPath 2.0 addressing over the 
DOM.  Can you address adequate performance using only the DOM interfaces 
as a data model?  Depending on what your primary goals are...

> - Should XSLT 2.0 be a clean room implementation?

So, what would your goals be for implementing XSLT 2.0 under the name of 
Xalan?  Is it simply that Saxon with schema support costs money?  I'm not 
sure that's enough of a driver.  Is it that you have a performance theory 
where you think you can do better than Saxon?  Is that you need 
functionality that you can not get from Saxon, and really feel you need to 
have full control?  Is it that you want a cleanly extensible XSLT 2.0 
implementation that people can build off of and freely experiment with?  I 
suspect this last is the most likely driver.

Taking that last as the primary driver, I might imagine an implementation 
over DOM, as simple and as clean as possible.

I could also imagine an implementation that only takes in XML text 
streams, and processes that in the most efficient possible form.

My main points are, you should have a damned good reason to spend a huge 
amount of work implementing the new features, and then be really centered 
on that purpose.

Also, XSLT 3.0 (http://www.w3.org/TR/xslt-30/) is in development.  Given 
the amount of time it would take to develop this, maybe that should be the 
actual target?

What about XQuery?  Do you eventually want to build an XQuery 
implementation from the core implementation?  Not saying this would be a 
bad or good thing, just something to consider.

You kind of open a Pandora's Box with considering further development in 
Xalan.  Could be a good thing, but I think you want to make certain you're 
clear on motivations and goals.  It's a lot of work.  (These are 
complicated specs.)

-scott




From:
Gary Gregory <[email protected]>
To:
"[email protected]" <[email protected]>, 
Date:
05/11/2014 10:55 AM
Subject:
Re: Xalan PMC / XLST 2 Support



Samuel,

I am glad to hear you are interesting in furthering Xalan development. I 
am interested in the Java side (I am a PMC member as well).

I do know that for the Java version, we do get asked from time to time 
when Xpath/XSLT 2.0 will be supported. The answer is usually along the 
lines of "when volunteers step in to do the work" ;)

We (I) just released Xalan-J 2.7.2 recently. The release process is old 
and clunky, and there is no one left it appears to help much. The 2.7.2 is 
only in the download/dist folder, I've not managed to release it to the 
Maven repo, nor have I updated the site... this will be all coming soon.

Another avenue to explore is:
- Should XSLT 2.0 be a clean room implementation?
- Should we try to reuse Eclipse's Psychopath XSLT/XPath 2.0 processor? It 
is even possible? Is licensing compatible?
- Should we try to reuse Saxon's processor? It is even possible? Is 
licensing compatible?

While dusting off the cobwebs, you should know that the Java SVN trunk is, 
AFAIK, in an unknown state, compared to the branch from which 2.7.2 was 
released.

You can scan the ML archives for my messages about building trunk vs. the 
2.7.2 branch.

Welcome aboard!
Gary


On Sun, May 11, 2014 at 8:42 AM, Samuel Queiroz <[email protected]> wrote:
Hi Xalan devs,

I am Samuel Queiroz and I worked with you in GSoC 12/13 and now I am in 
the process of being a Xalan PMC member. I am glad for this!

I plan to put my efforts on implementing full XSLT 2 support on Xalan.

For this, I would like to know which version, Java or C/C++, is more 
advanced in terms of features.
Which one is more used? Which one the users claim for XSLT 2.0 support? 

After defining where I am going to put my efforts, my plan is as follows:
1.      To study the project architecture and to become a XSLT 1.0 expert;
2.      To do some code refactoring, if necessary;
3.      To write a simple but complete Xalan developer architecture 
documentation, in order to help new Xalan developers;
4.      Create more tests;

At this point, the project will be really well tested and well implemented 
(I am not saying it is not; I am just planning to do this if necessary).

5.      To create a plan to have XSLT 2.0 support ...
What do you think about it?

Sincerely,
Samuel



-- 
E-Mail: [email protected] | [email protected] 
Java Persistence with Hibernate, Second Edition
JUnit in Action, Second Edition
Spring Batch in Action
Blog: http://garygregory.wordpress.com 
Home: http://garygregory.com/
Tweet! http://twitter.com/GaryGregory

Reply via email to