Re: In PG12, query with float calculations is slower than PG11 - Mailing list pgsql-hackers

From Tom Lane
Subject Re: In PG12, query with float calculations is slower than PG11
Date
Msg-id 9367.1581615731@sss.pgh.pa.us
Whole thread Raw
In response to Re: In PG12, query with float calculations is slower than PG11  (Andres Freund <andres@anarazel.de>)
List pgsql-hackers
Andres Freund <andres@anarazel.de> writes:
> On February 13, 2020 8:30:45 AM PST, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> I see some minor things I don't like here, eg float_*flow_error()
>> need some documentation as to why they exist.  But I'll review,
>> fix those things up and then push.

> Would be good to mark them noreturn too.

Yeah, that was one of the things I didn't like ;-).  Also the lack
of pg_noinline.

> Wonder if it's useful to add the"cold" marker to pg. Not as part of this patch, but for functions like these.

I'm only seeing about a 1.5kB reduction in the backend size from
this patch, which kinda surprises me, but it says that we're
not winning all that much from just having one copy of the ereport
calls.  So I don't think that "cold" is going to add much.

            regards, tom lane



pgsql-hackers by date:

Previous
From: Andres Freund
Date:
Subject: Re: In PG12, query with float calculations is slower than PG11
Next
From: Tom Lane
Date:
Subject: Re: In PG12, query with float calculations is slower than PG11