From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Andres Freund <andres(at)anarazel(dot)de> |
Cc: | Robert Haas <robertmhaas(at)gmail(dot)com>, Jeevan Ladhe <jeevan(dot)ladhe(at)enterprisedb(dot)com>, PostgreSQL Developers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Binary search in fmgr_isbuiltin() is a bottleneck. |
Date: | 2017-09-28 23:06:27 |
Message-ID: | 15644.1506639987@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Andres Freund <andres(at)anarazel(dot)de> writes:
> On 2017-09-28 18:52:28 -0400, Tom Lane wrote:
>> Uh, what? Access to fmgr_nbuiltins shouldn't be part of any critical path
>> anymore after this change.
> Indeed. But the size of the the oid -> fmgr_builtins index array is
> relevant now. We could of course just make that dependent on
> FirstBootstrapObjectId, but that'd waste some memory.
Not enough to notice, considering there are pg_proc OIDs up in the 8K
range already. We blow 2KB of never-accessed space for far less good
reason than this.
>> I'm kind of -0.5 on that. I believe part of the argument for having
>> things set up as they were was to allow external code to access the
>> fmgr_builtins table (as my speed-test hack earlier today did).
> You could still do that, you'd just end up with a second copy. Doesn't
> seem bad for such an uncommon case.
If I understand what you're proposing, it would involve the extension
containing its *own* copy of the fmgr table, which seems pretty horrid.
It wouldn't necessarily match the actual contents in the core executable.
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Tom Lane | 2017-09-29 00:01:37 | Re: Minor codegen silliness in ExecInterpExpr() |
Previous Message | Alexander Korotkov | 2017-09-28 22:59:01 | Re: GSoC 2017: weekly progress reports (week 8) |