specify the REGION function with reference to DALI shape

[pdowler]

Once DALI specifies xtype="shape" this provides a replacement for the non-normative usage of "adql:region" from TAP-1.0 with something that is a normative part of a standard, with most of the features intact (polymorphism) and the sketchy bits removed (coordsys that could change from row to row).

This will be in DALI-1.2 (subtle details close to worked out) so we could consider it for later, but ADQL-next could be a long way off so I would prefer get it into ADQL-2.1 if we can.

[msdemlei]

On Wed, Jun 07, 2023 at 01:01:04PM -0700, Patrick Dowler wrote: This will be in DALI-1.2 (subtle details close to worked out) so we could consider it for later, but ADQL-next could be a long way off so I would prefer get it into ADQL-2.1 if we can.

...but let's not delay 2.1 any more because of this either and rather try and get 2.2 out of the door a lot faster; but then I don't expect ADQL will go to REC before the summer either.

[gmantele]

Sorry for answering so late...last weeks were so crazy 🙃

Thank you very much @pdowler for this good piece of news :-) It is indeed really interesting for, at least, Obscore+TAP.

But, as much as I would like to include shape and MOC inside ADQL, I agree with Markus: this should not delay ADQL-2.1 which made so much time to finally be released. I also agree that next versions of ADQL should be released much faster than that, and especially ADQL-2.2 with PEG and MOC. I think it is ok (and even preferable) that DALI releases the shape and MOC xtypes before ADQL-2.2 uses them. In the meantime, nothing prevents TAP service providers to use them in their service. Besides, it will take also some time too for Obscore to adapt its specification to this new shape ; in a way, releasing ADQL and then Obscore so that following the new DALI-REC seems perfectly ok to me.

[pdowler]

OK, that sounds reasonable. Please mark this as "enhancement" for a future version.