ADQL-2.1's status report

[Gregory MANTELET]

Hello DAL,

Here is a little status report about the ADQL-2.1 document.

We already have some open issues/Pull-Requests that would need a review.
If some of the following topics interest you, do not hesitate to take a
look on the corresponding Issue/PR description and discussion.

* * Open issues:*

    - Boolean <https://github.com/ivoa-std/ADQL/issues/32>
    - UDF-catalogue <https://github.com/ivoa-std/ADQL/issues/16>
    - Geometries fix <https://github.com/ivoa-std/ADQL/issues/29>
    - Coordinate system specification 
<https://github.com/ivoa-std/ADQL/issues/10>

* * Pull Requests under review:*

    - UDF-catalogue <https://github.com/ivoa-std/ADQL/pull/26>
    - Geometries fix <https://github.com/ivoa-std/ADQL/pull/41>

* * Brief summaries:*

*- Boolean:*
      It has been proposed to add the boolean values `True` and `False` 
and to
      allow boolean comparison with resp. 1/0 and True/False. It would 
also be
      possible to write boolean expressions as such without a comparison.
      Considering the non-homogeneous aspect of existing DBMS regarding the
      boolean values/datatype and the complexity raise of the grammar if 
we want
      to support it, the current decision is to postpone `boolean` for a 
later
      version of ADQL

/*=> Go see the discussion for the other proposed alternatives.*//*
*//*        In absence of answer, ADQL-2.1 will not include any boolean 
for now.*/

*- UDF-catalogue:*
      An Endorsed Note gathering the definitions of frequently used User 
Defined
      Functions. The idea is that everybody uses the same definition, 
for the
      sake of our users. This EN should be soon accepted.

/*=> At that moment, the corresponding Issue and Pull-Request will be*//*
*//*        completed in ivoa-std/ADQL. No input required for this topic.*/

*- Geometries Fix:*
      This fix aims to fix a syntax ambiguous situation with `POLYGON` 
due to the
      first argument (coordinate system) becoming optional. This has 
finally been
      fixed by a new Erratum about the datatype of the coordinate system
      argument.

/*=> The corresponding Pull-Request is waiting for someone to 
review/accept*//*
*//*        it. After that, it and its corresponding Issue can be closed.*/

*- Coordinate system specification:*
      With ADQL-2.1, the coordinate system argument of geometries like 
`POINT`,
      `CIRCLE`, ... become deprecated and so, it is optional. In a later 
version,
      it will be removed. It means that it is no longer possible to 
explicitly
      specify the input coordinate system of geometries. The consequence 
is that
      coordinate system convertion is not explicitely possible any more 
while
      trying to match geometries each other (e.g. with `CONTAINS`, 
`INTERSECTS`)
      (note that nothing prevents the ADQL service to perform it 
automatically
      with an internal mechanism). This is a feature actually supported 
by very
      few TAP services but is however very important for services offering
      positions at different coordinate systems (e.g. GAVO, VizieR). The 
proposed
      solution is to rely on UDF-Catalogue. GAVO has already defined 
some UDFs
      which are listed in this PEN, but with the prefix `gavo_`. When 
other TAP
      services will need such functions, there should be discussions all 
together
      to converge toward a common UDF definition ; once done, these 
functions
      will be prefixed with `ivo_`.

/*=> I propose to close this issue unless there are other reactions.*/


* * Remaining issues:*

    - Fix set operators <https://github.com/ivoa-std/ADQL/issues/20>
    - TIMESTAMP vs CAST <https://github.com/ivoa-std/ADQL/issues/11>


Do not hesitate to reactin the discussionson GitHub, or otherwise by email.

Cheers,
Grégory M.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ivoa.net/pipermail/dal/attachments/20201016/14e0442b/attachment.html>