Re: Allow to collect statistics on virtual generated columns
Dean Rasheed <dean.a.rasheed@gmail.com>
From: Dean Rasheed <dean.a.rasheed@gmail.com>
To: Yugo Nagata <nagata@sraoss.co.jp>
Cc: Andres Freund <andres@anarazel.de>, pgsql-hackers@postgresql.org
Date: 2026-04-01T16:12:38Z
Lists: pgsql-hackers
On Tue, 31 Mar 2026 at 15:35, Yugo Nagata <nagata@sraoss.co.jp> wrote: > > Thank you for updating the patch. I am fine with that. > > One concern is that users might interpret "stored" as referring to > "stored generated columns", rather than including regular columns, Yeah, possibly. I changed "stored" to "non-virtual", which should reduce the chances of that particular confusion. > Also, the meaning of "automatically" might be a bit unclear, so we > could clarify it by adding "without defining extended statistics." > > Defining extended statistics on a single <emphasis>stored</emphasis> > column is not supported or necessary, because statistics are built > automatically on such columns without defining extended statistics. OK, pushed that way. I also noticed that a few places in CreateStatistics() could use the variable "numcols" instead of "list_length(stmt->exprs)", so I changed that. I decided to include the change to the error message discussed in [1], since there seemed to be a consensus there, except that I think it also needs to make it clear that it refers only to non-virtual columns. In addition, there was another nearby error message which was no longer quite right for statistics on virtual generated columns, and I changed the order of checks, since that allowed the "if" statements to be simplified a bit. Regards, Dean [1] https://www.postgresql.org/message-id/20260324203430.411331c59ca462457ec0aa8b%40sraoss.co.jp
Commits
-
Add support for extended statistics on virtual generated columns.
- f7f4052a4e95 19 (unreleased) landed