From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Chris Bandy <bandy(dot)chris(at)gmail(dot)com> |
Cc: | Adam Brusselback <adambrusselback(at)gmail(dot)com>, Michael Paquier <michael(dot)paquier(at)gmail(dot)com>, Paul A Jungwirth <pj(at)illuminatedcomputing(dot)com>, PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: GiST support for UUIDs |
Date: | 2016-11-29 17:58:42 |
Message-ID: | 1407.1480442322@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Chris Bandy <bandy(dot)chris(at)gmail(dot)com> writes:
> [ btree_gist_uuid_8.patch ]
Um ... is there a reason why the penalty logic in gbt_uuid_penalty()
is completely unlike that for any other btree_gist module?
As best I can tell from the (admittedly documentation-free) code
elsewhere, the penalty ought to be proportional to the fraction
by which the original range is expanded; that's not what this
code is doing. It also seems to be missing the machinations related
to scaling per-column results in a multi-column index.
I'm kind of inclined to change uuid_parts_distance to just convert
a given pg_uuid_t to "double" and then apply penalty_num(), as is
done in gbt_macad_penalty.
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Jeff Janes | 2016-11-29 17:58:50 | Re: Time to up bgwriter_lru_maxpages? |
Previous Message | Robert Haas | 2016-11-29 17:51:09 | Re: Proposal: scan key push down to heap [WIP] |