XSLT 2.0 processor discussion

[Mukul Gandhi <ga...@gmail.com>]

Hi all,
   I've been tempted to bring this topic again, to this list. I think it'd
be really great for Xalan to have a XSLT 2.0 processor as well (its 1.0
processor is just great). I'll attempt to enumerate following options with
my notions of pros-and-cons, for Xalan to provide an XSLT 2.0 processor as
well,
1) Natively enhance Xalan's current XSLT 1.0 processor to a 2.0 processor,
in a branch of its own. The effort and time to market for this, would be
highest.
2) Use Eclipse's XPath 2.0 processor (also known as PsychoPath XPath 2.0
processor), in the XSLT 2.0 codebase which Xalan would develop. This option
saves us the effort of writing a native XPath 2.0 processor. But perhaps
the cons is that, PsychoPath XPath 2.0 processor can accept only the DOM
input to convert it to XDM.
3) Is there any possibility of IBM donating in some way its XSLT 2.0
processor technology to Xalan? If it can be done, people will flock to the
IBM derived Xalan's XSLT 2.0 processor.
4) Can we use Saxon's latest home edition XSLT 2.0 processor (its open
source), and convert to to Xalan's XSLT 2.0 processor? I believe, Saxon's
open source products come with Mozilla Public License, and I'm not sure how
suitable it is for having it in Xalan?

I'd also suggest, that any XSLT 2.0 processor from Xalan should be
schema-aware using Xerces's XSD processor.

I'm still not thinking about XSLT 3.0, which is already a W3C spec. I think
we should have XSLT 2.0 first.

Needless to say, I'll be happy to participate in any such work.



-- 
Regards,
Mukul Gandhi

Re: XSLT 2.0 processor discussion

[Mukul Gandhi <ga...@gmail.com>]

On Wed, May 30, 2018 at 2:36 AM, Scott Boag <sc...@us.ibm.com> wrote:

>
> In order to build that community, a strong purpose would need to drive it,
> above what the current os offerings provide, including Saxon.  So, give it
> some thought.
>

my personal thoughts,

I think despite IBM not willing to support further XSLT (3.0) development
at Xalan, we should still have a goal to implement XSLT 3.0 at Xalan. Yes
and I strongly feel (agreeing to Scott's point earlier) that we should
start developing for XSLT 3.0 and not 2.0 (we can skip XSLT 2.0 development
altogether). We should implement the 3.0 processor, as schema aware,
supporting streaming and with all other goodies defined in the spec.

In case, Xalan decides to start developing XSLT 3.0 processor, I'll be
happy to be a committer to that project (I'm also a Xerces-J committer);
sorry if this sounds rude. I'd like to contribute actively to any XSLT 3.0
development at Xalan.



-- 
Regards,
Mukul Gandhi

Re: XSLT 2.0 processor discussion

[Gary Gregory <ga...@gmail.com>]

Hi All:

I like all this talk of XSLT 2 and 3. I applied a couple of patches years
ago and released 2.7.2, but there has been very little activity on this
project before and after that.

There has been recent activity in JIRA regarding patches and reports, but I
have not taken the time to research or apply them. I am by no means a Xalan
expert here. When I asked for feedback a couple of years ago on this ML
regarding the state of various branches, I got zero replies IIRC.

So we might be facing a steep learning curve here, what amounts to
effectively resuscitating this valuable project.

Gary

On Mon, May 28, 2018 at 11:34 PM, Mukul Gandhi <ga...@gmail.com>
wrote:

> Hi Scott,
>    Its really nice to see your reply. Please see my comments to your
> points, below
>
> On Mon, May 28, 2018 at 9:20 PM, Scott Boag <sc...@us.ibm.com> wrote:
>>
>>
>> > xslt20 & xslt20-compiled
>> >
>> > Any thoughts, what is the functionality currently contained in above
>> folders?
>>
>> I think those were *very* early attempts to do XSLT 2.0 on Xalan?  Would
>> be interesting to get dates on these.
>>
>
> At this link, http://svn.apache.org/viewvc/xalan/java/branches/ its
> mentioned these folders had previous edits 14 yrs ago.
>
>
>> > Is there any possibility of IBM donating in some way its XSLT 2.0
>> processor technology to Xalan? If it can be done, people will flock to the
>> IBM derived Xalan's XSLT 2.0 processor.
>>
>> I think it could be a possibility.  Not sure it's wise or not... have to
>> think about it a bit. But let me discuss with some folks in IBM to explore
>> the options.
>>
>
> Thanks. IBM donated its XML Parsing & API technology earlier to apache
> (Xerces. The Java one I guess). I feel, same could be done with XSLT 2.0.
> It would be just great, and beneficial both to community and of course IBM.
>
> > Can we use Saxon's latest home edition XSLT 2.0 processor (its open
>> source), and convert to to Xalan's XSLT 2.0 processor?
>>
>> Hmm, seems on the surface to be a non-starter to me, just given it does
>> not feel right.
>>
>
> I quite agree with you.
>
>
>> I was writing some XSLT the other day, and was reminded what a cool and
>> interesting language it is
>>
>
> I agree.
>
>
>> despite it's somewhat tarnished reputation.
>>
>
> To my opinion, XSLT is a great language in a particular niche (i.e XML
> transformation). Its not a general purpose language, but yet its Turing
> complete I believe.
>
>
>> To justify the work for a new processor, really there should be a core
>> idea that makes it potentially unique and valuable, above the current state
>> of the art.
>>
>
> I guess, if you could jump into new XSLT development (2.0 & 3.0) at Xalan,
> we'll gain a lot from your leadership.
>
>
>> For myself, I'd love to see a 3.0 processor implemented in Typescript for
>> client use
>>
>
> Targeting Xalan to have XSLT 3.0 instead of 2.0 is a good idea. I'm not
> much aware of Typescript, so can't comment on that. I guess, implementing
> Xalan's XSLT 3.0 in Java would be just great.
>
>
>
>> Still, given the ability to do JSON transformations, and mix
>> XML/JSON/XHTML, and throwing in the ability to do XQuery 3.0, it could be
>> really exciting, from a functional viewpoint.
>>
>
> I agree.
>
>
>
> --
> Regards,
> Mukul Gandhi
>

Re: XSLT 2.0 processor discussion

[Mukul Gandhi <ga...@gmail.com>]

Hi Scott,
   Its really nice to see your reply. Please see my comments to your
points, below

On Mon, May 28, 2018 at 9:20 PM, Scott Boag <sc...@us.ibm.com> wrote:
>
>
> > xslt20 & xslt20-compiled
> >
> > Any thoughts, what is the functionality currently contained in above
> folders?
>
> I think those were *very* early attempts to do XSLT 2.0 on Xalan?  Would
> be interesting to get dates on these.
>

At this link, http://svn.apache.org/viewvc/xalan/java/branches/ its
mentioned these folders had previous edits 14 yrs ago.


> > Is there any possibility of IBM donating in some way its XSLT 2.0
> processor technology to Xalan? If it can be done, people will flock to the
> IBM derived Xalan's XSLT 2.0 processor.
>
> I think it could be a possibility.  Not sure it's wise or not... have to
> think about it a bit. But let me discuss with some folks in IBM to explore
> the options.
>

Thanks. IBM donated its XML Parsing & API technology earlier to apache
(Xerces. The Java one I guess). I feel, same could be done with XSLT 2.0.
It would be just great, and beneficial both to community and of course IBM.

> Can we use Saxon's latest home edition XSLT 2.0 processor (its open
> source), and convert to to Xalan's XSLT 2.0 processor?
>
> Hmm, seems on the surface to be a non-starter to me, just given it does
> not feel right.
>

I quite agree with you.


> I was writing some XSLT the other day, and was reminded what a cool and
> interesting language it is
>

I agree.


> despite it's somewhat tarnished reputation.
>

To my opinion, XSLT is a great language in a particular niche (i.e XML
transformation). Its not a general purpose language, but yet its Turing
complete I believe.


> To justify the work for a new processor, really there should be a core
> idea that makes it potentially unique and valuable, above the current state
> of the art.
>

I guess, if you could jump into new XSLT development (2.0 & 3.0) at Xalan,
we'll gain a lot from your leadership.


> For myself, I'd love to see a 3.0 processor implemented in Typescript for
> client use
>

Targeting Xalan to have XSLT 3.0 instead of 2.0 is a good idea. I'm not
much aware of Typescript, so can't comment on that. I guess, implementing
Xalan's XSLT 3.0 in Java would be just great.



> Still, given the ability to do JSON transformations, and mix
> XML/JSON/XHTML, and throwing in the ability to do XQuery 3.0, it could be
> really exciting, from a functional viewpoint.
>

I agree.



-- 
Regards,
Mukul Gandhi

Re: XSLT 2.0 processor discussion

[Mukul Gandhi <ga...@gmail.com>]

Interestingly, I can see at this location
http://svn.apache.org/repos/asf/xalan/java/branches/, following folders

xslt20 & xslt20-compiled

Any thoughts, what is the functionality currently contained in above
folders?

On Sat, May 26, 2018 at 11:41 AM, Mukul Gandhi <ga...@gmail.com>
wrote:

> Hi all,
>    I've been tempted to bring this topic again, to this list. I think it'd
> be really great for Xalan to have a XSLT 2.0 processor as well (its 1.0
> processor is just great). I'll attempt to enumerate following options with
> my notions of pros-and-cons, for Xalan to provide an XSLT 2.0 processor as
> well,
> 1) Natively enhance Xalan's current XSLT 1.0 processor to a 2.0 processor,
> in a branch of its own. The effort and time to market for this, would be
> highest.
> 2) Use Eclipse's XPath 2.0 processor (also known as PsychoPath XPath 2.0
> processor), in the XSLT 2.0 codebase which Xalan would develop. This option
> saves us the effort of writing a native XPath 2.0 processor. But perhaps
> the cons is that, PsychoPath XPath 2.0 processor can accept only the DOM
> input to convert it to XDM.
> 3) Is there any possibility of IBM donating in some way its XSLT 2.0
> processor technology to Xalan? If it can be done, people will flock to the
> IBM derived Xalan's XSLT 2.0 processor.
> 4) Can we use Saxon's latest home edition XSLT 2.0 processor (its open
> source), and convert to to Xalan's XSLT 2.0 processor? I believe, Saxon's
> open source products come with Mozilla Public License, and I'm not sure how
> suitable it is for having it in Xalan?
>
> I'd also suggest, that any XSLT 2.0 processor from Xalan should be
> schema-aware using Xerces's XSD processor.
>
> I'm still not thinking about XSLT 3.0, which is already a W3C spec. I
> think we should have XSLT 2.0 first.
>
> Needless to say, I'll be happy to participate in any such work.
>




-- 
Regards,
Mukul Gandhi