From: | Simon Riggs <simon(at)2ndQuadrant(dot)com> |
---|---|
To: | Robert Haas <robertmhaas(at)gmail(dot)com> |
Cc: | PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Deprecating Hash Indexes |
Date: | 2012-10-15 17:26:43 |
Message-ID: | CA+U5nMJTuFCBWU-WB0rbo7h=XUrqMjw8gsLO4Tm1gFOdYsh_FQ@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On 15 October 2012 18:07, Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
> On Mon, Oct 15, 2012 at 12:59 PM, Simon Riggs <simon(at)2ndquadrant(dot)com> wrote:
>>> I don't think I'd go so far as to say that we should
>>> imply that they'll be removed in a future release. Given how deeply
>>> intertwined they are with the planner, I doubt that that will happen;
>>> and I think there is enough interest in the technology that it's
>>> likely to eventually be fixed.
>>
>> Hash indexes aren't used in the planner. Hash joins use completely
>> separate code.
>
> It's not really completely separate, because to do a hash join we have
> to find a hash function for the relevant data types, and IIUC we do
> that by looking up the default hash opclass for the datatype and
> finding its first support function. Of course, if we were to remove
> the hash AM, then you couldn't define a hash opclass against it.
Presumably it defaults to hash_any() but I get the picture.
--
Simon Riggs http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services
From | Date | Subject | |
---|---|---|---|
Next Message | ktm@rice.edu | 2012-10-15 17:36:06 | Re: Deprecating Hash Indexes |
Previous Message | Josh Berkus | 2012-10-15 17:21:15 | Re: Potential autovacuum optimization: new tables |