ADQL-2.1 internal draft

Dave Morris dave.morris at metagrid.co.uk
Fri Jun 12 13:03:13 CEST 2015


Hi Walter,

On 2015-06-12 08:29, Walter Landry wrote:
> 
> Do you have a list of what works on which platforms?  I would be very
> curious to see it.
> 

Unfortunately no, we don't have a definitive list, which is part of the 
problem.

I did a check when I proposed the bitwise operations and found that 
Oracle, Derby and SQLite had issues with some of the operations.
https://volute.googlecode.com/svn/trunk/projects/dal/TAPNotes/TAPNotes-fmt.html#af-bitwise

When editing the text for the rest of the specification I did a quick 
Google search for each feature and it looks like there may be issues 
with the set operators on MySQL.

WITH
https://stackoverflow.com/questions/1382573/how-do-you-use-the-with-clause-in-mysql
https://bugs.mysql.com/bug.php?id=23156

EXCEPT
https://stackoverflow.com/questions/13019865/except-all-equivalent-in-mysql
http://forums.mysql.com/read.php?10,513537,513537#msg-513537

INTERSECT
https://stackoverflow.com/questions/2621382/alternative-to-intersect-in-mysql

Some of these issues may be solved by parsing and rewriting the query - 
hence the proposal to drop the 'no rewrite' requirement.

However, this is not an exhaustive search, it is not covering all the 
platforms and it is definitely not a strict verification.

"Best guess based on what Dave found with Google" is probably not a 
valid criteria for accepting or rejecting mandatory features.

I erred on the side of caution this time, mainly because in several 
cases I knew enough to know I don't know what the answer is.

If I found something that suggested there was a problem with 
implementing a feature, then it made sense to make it optional in this 
version. If we solve the issues, there is no problem making them 
mandatory in the next version. Going the other way mandatory -> optional 
is more problematic.

Cheers,
Dave


--------
Dave Morris
Senior Troll
Wide Field Astronomy Unit
Institute for Astronomy
University of Edinburgh
--------



More information about the dal mailing list