W3C home > Mailing lists > Public > xproc-dev@w3.org > July 2020

Re: Node identity in XProc 3

From: Achim Berndzen <achim.berndzen@xml-project.com>
Date: Wed, 8 Jul 2020 15:34:40 +0200
Cc: XProc Dev <xproc-dev@w3.org>, Norm Walsh <ndw@nwalsh.com>
Message-Id: <FB52115D-20F5-47F0-9C11-45631659D299@xml-project.com>
To: Martin Honnen <martin.honnen@gmx.de>

> Am 08.07.2020 um 14:38 schrieb Norman Tovey-Walsh <ndw@nwalsh.com>:
> Martin Honnen <martin.honnen@gmx.de> writes:
>> I am not sure whether that is supposed to be preserved when the array is
>> the result of a p:xquery step.
>> Thoughts?
> With the possible exception of p:identity (and maybe a couple of other
> steps), node identity should not be expected.
>                                        Be seeing you,
>                                          norm

I completely agree with Norms analysis. 
If we would stipulate node identity say for the XQuery in question or an equivalent XSLT, this would result in a massive vendor lock: The node model used in the XProc implementation, as well as in p:xquery and p:xslt would need to be the same. Any conversion what so ever will al least threaten node identity, so no conversion could take place.

The second point is that it would dramatically reduce the number of available tools as basis for step implementation. I could not choose the tool which (I think) is best for users, but I had to choose a tool which is able to preserve node identity. So focusing on node identity would effectivly result in inferior XProc implementation.

Received on Wednesday, 8 July 2020 13:35:00 UTC

This archive was generated by hypermail 2.4.0 : Wednesday, 8 July 2020 13:35:01 UTC