On 10/27/20 7:11 AM, Ibrar Ahmed wrote:
> I have spent some more time on the patch and did a lot of cleanup along
> with some fixes, compilation errors, and warnings.
Thank you for taking a look at this! I've been swamped with ordinary
work and haven't had a chance to focus on it for a while, but I'm hoping
to make some improvements over the coming holidays, especially based on
feedback from my talk at PgCon. There are a handful of small specific
things I'd like to do, and then one big thing: add support for PERIODs.
Vik said I could include his old patch for PERIODs, so I'd like to get
that working on the latest master, and then rebase my own work on top of
it. Then we can accept either ranges or PERIODs in various places
(marked by TODOs in the code).
Vik also pointed out a way to check foreign keys without using
range_agg. He thinks it may even be more efficient. On the other hand
it's a much more complicated SQL statement. I'd like to do a performance
comparison to get concrete numbers, but if we did use his query, then
this patch wouldn't depend on multiranges anymore---which seems like a
big aid to moving it forward. Assuming multiranges gets committed, we
can always swap in the range_agg query depending on the performance
comparison results.
I apologize for the slow progress here, and thank you for your help!
Yours,
--
Paul ~{:-)
pj@illuminatedcomputing.com