Horton

Re: Bogus collation version recording in recordMultipleDependencies

Julien Rouhaud <rjuju123@gmail.com>

View thread View on postgr.es
From: Julien Rouhaud <rjuju123@gmail.com>
To: Peter Geoghegan <pg@bowt.ie>
Cc: Thomas Munro <thomas.munro@gmail.com>, Tom Lane <tgl@sss.pgh.pa.us>, Andres Freund <andres@anarazel.de>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2021-04-20T01:46:07Z
Lists: pgsql-hackers 
On Mon, Apr 19, 2021 at 11:13:37AM -0700, Peter Geoghegan wrote:
> On Sun, Apr 18, 2021 at 4:23 AM Julien Rouhaud <rjuju123@gmail.com> wrote:
> > So IIUC the issue here is that the code could previously record useless
> > collation version dependencies in somes cases, which could lead to false
> > positive possible corruption messages (and of course additional bloat on
> > pg_depend).  False positive messages can't be avoided anyway, as a collation
> > version update may not corrupt the actually indexed set of data, especially for
> > glibc.
> 
> This argument seems completely absurd to me.

I'm not sure why?  For glibc at least, I don't see how we could not end up
raising false positive as you have a single glibc version for all its
collations.  If a user has say en_US and fr_FR, or any quite stable collation,
most of the glibc upgrades (except 2.28 of course) won't corrupt your indexes.

Commits

  1. Doc: Update notes about libc collation versions.

  2. Revert per-index collation version tracking feature.

  3. Rethink extraction of collation dependencies.

  4. Fix bogus collation-version-recording logic.

  5. Rename the "point is strictly above/below point" comparison operators.

  6. Move catalog toast table declarations