adding Xalan's XSL 3 implementation within jdk

[Joe Wang]

Hi Mukul,

Xalan-J's XSL 3 implementation has not been integrated into any Xalan 
releases. I understand it implemented most of the XSLT 3.0 
specification, but not all of them, esp. the Streaming feature. Was the 
incompleteness the reason why it was not included in an Xalan release? 
What's your assessment on the readiness for a formal release (or how 
much additional work is needed)? What are the conformance test results?

Also, do you have data showing how the Xalan-J's XSL 3 implementation is 
used in user applications? What are the feedback (or bug reports) from 
developers?

Thanks,

Joe


On 2/26/25 7:59 AM, Mukul Gandhi wrote:
> Hi Alan,
>      I've just seen this mail from you. Apologies for a delayed response.
>
> My mail box has had few issues due to the volume of mails that I get
> from mailing lists.
>
> On Sun, Feb 2, 2025 at 9:38 PM Alan Bateman <alan.bateman at oracle.com> wrote:
>
>> The stats for that branch suggest 5,845 changed files with 234,372 additions and 84,058 deletions. I can't easily tell how much of this would need to come into the jdk repo but this looks like a major update. If only 10% of this is applicable to the JDK then it still needs seems like a major update that would require a huge investment to audit and integrate this code. How much XML is in new applications developed in 2025? Only asking because it's an area that is surely much lower priority compared to all the other major investments right now. Maybe there are useful security or performance changes that would be useful to cherry pick instead? Finally, does this Xalan update work with the SPIs so that someone really looking for XSL 3 can just deploy it on the class path and module path?
> Ofcourse, anyone could use Xalan-J's XSL 3 implementation with JDK by
> placing Xalan jars on class path & module path.
>
> Since Xalan-J's XSLT 1.0 & XPath 1.0 implementations are already
> available within JDK, I thought its natural if JDK could pick
> Xalan-J's XSL 3 implementation and include that within JDK. I can
> imagine that this may surely be time consuming for someone from JDK
> team to integrate with JDK. XSLT 1.0's use I think is very less these
> days particularly for new XML projects, due to vast improvements in
> language features offered by XSLT 3.0 and XPath 3.1.
>
> IMHO, I wrote all the XSL 3 implementation code (and solved various
> XSL 3 implementation bugs reported by community on Xalan-J's dev
> forum) within Xalan-J's XSL 3 dev respos branch, enhancing upon
> Xalan-J's XSLT 1.0 implementation. From my point of view, I'll be
> happy if JDK could include Xalan-J's XSL 3 implementation.
>
> I even wrote following two online articles on xml.com about few of XSL
> 3 language features, and how they're implemented within Xalan-J,
> https://www.xml.com/articles/2024/07/22/string-analysis-with-analyze-string/
> https://www.xml.com/articles/2023/12/05/xml-path-language-xpath-higher-order-functions/
>
>
> Many thanks.
>
>