Re: Request for information about postgres version 15.2 stability - Mailing list pgsql-general

From Tom Lane
Subject Re: Request for information about postgres version 15.2 stability
Date
Msg-id 2664328.1681835752@sss.pgh.pa.us
Whole thread Raw
In response to Re: Request for information about postgres version 15.2 stability  (Laurenz Albe <laurenz.albe@cybertec.at>)
List pgsql-general
Laurenz Albe <laurenz.albe@cybertec.at> writes:
> We don't have any information about yet undiscovered PostgreSQL bugs.

Indeed.

> We believe that PostgreSQL 15.2 is a great and stable database, but we are biased.

One moderately objective way to see how stable a release branch is
is to count how many commits fix bugs in that branch but not any
older branch.  If a bug would be problematic for you but it also
appears in (say) v14 and v13, then choosing to use one of those
branches instead of v15 wouldn't have saved you.  Also, if a commit
goes back that far, it means that the bug escaped detection for
multiple years in production, which suggests that it is minor and/or
doesn't affect many people.  (I hasten to add that we do find some
bugs that are both old and serious.)

A quick review of commits since 15.2 says that practically all of the
ones that don't also affect v14 or before have to do with glitches in
new-in-15 features, such as the MERGE command.  Which is code you
wouldn't be using anyway if your application would be satisfied with
an older branch.  So I'd judge that 15 is pretty stable at this point,
and there is very little reason to prefer any older branch.

If you want to do your own analysis along these lines, the gitweb
interface isn't especially friendly for comparing commit logs of
different branches.  I'd suggest using our git_changelog script:

git clone https://git.postgresql.org/git/postgresql.git
cd postgresql
src/tools/git_changelog >../pg-revision-history

This produces output that collates matching commits in different
branches.

            regards, tom lane



pgsql-general by date:

Previous
From: Laurenz Albe
Date:
Subject: Re: Logical Replication Sync Validation
Next
From: Inzamam Shafiq
Date:
Subject: ETCD for Patroni configuration error