Horton

Re: Patch: shouldn't timezone(text, timestamp[tz]) be STABLE?

John Naylor <john.naylor@enterprisedb.com>

View thread View on postgr.es
From: John Naylor <john.naylor@enterprisedb.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Aleksander Alekseev <aleksander@timescale.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2021-08-31T17:34:06Z
Lists: pgsql-hackers 
On Mon, Aug 30, 2021 at 12:58 PM Tom Lane <tgl@sss.pgh.pa.us> wrote:
>
> Aleksander Alekseev <aleksander@timescale.com> writes:
> > Got it. But in this case, what's your opinion on the differences between
> > date_trunc() and timezone()? Shouldn't date_trunc() be always IMMUTABLE
as
> > well?
>
> No, because date_trunc depends on the current timezone setting,
> or at least its stable variants do.

A light bulb went off in my head just now, because I modeled date_bin() in
part on date_trunc(), but apparently it didn't get the memo that the
variant with timezone should have been marked stable.

I believe it's been discussed before that it'd be safer if pg_proc.dat had
the same defaults as CREATE FUNCTION, and this is further evidence for that.

--
John Naylor
EDB: http://www.enterprisedb.com

Commits

  1. Fix actively-misleading comments about the contents of struct pg_tm.

  2. Fix bogus timetz_zone() results for DYNTZ abbreviations.

  3. Make timetz_zone() stable, and correct a bug for DYNTZ abbreviations.