Re: Missing CFI in hlCover()? - Mailing list pgsql-hackers

From Stephen Frost
Subject Re: Missing CFI in hlCover()?
Date
Msg-id 20200731123625.GB12375@tamriel.snowman.net
Whole thread Raw
In response to Re: Missing CFI in hlCover()?  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-hackers
Greetings,

* Tom Lane (tgl@sss.pgh.pa.us) wrote:
> Stephen Frost <sfrost@snowman.net> writes:
> > * Tom Lane (tgl@sss.pgh.pa.us) wrote:
> >> Here's a proposed patch along that line.
>
> > I've back-patched this to 11 (which was just a bit of fuzz) and tested
> > it out with a couple of different queries that were causing issues
> > previously on the archive server, and they finish in a much more
> > reasonable time and react faster to cancel requests/signals.
>
> Yeah, I'd tried this locally using the data from the one test case you
> showed me, and it seemed to fix that.

Good stuff.

> > So, looks good to me, and would certainly be nice to get this into the
> > next set of releases, so the archive server doesn't get stuck anymore.
>
> I'll push this tomorrow if nobody has objected to it.

Sounds good.

> BTW, I had noticed last night that hlFirstIndex is being unreasonably
> stupid.  Many of the "words" have null item pointers and hence can't
> possibly match any query item (I think because we have "words" for
> inter-word spaces/punctuation as well as the actual words).  Checking
> that, as in the attached v2 patch, makes things a bit faster yet.

Nice, looks good to me.

Thanks!

Stephen

Attachment

pgsql-hackers by date:

Previous
From: Ashutosh Sharma
Date:
Subject: Re: recovering from "found xmin ... from before relfrozenxid ..."
Next
From: Hamid Akhtar
Date:
Subject: Re: track_planning causing performance regression