Re: [HACKERS] removing abstime, reltime, tinterval.c, spi/timetravel

Andres Freund <andres@anarazel.de>

From: Andres Freund <andres@anarazel.de>
To: Mark Dilger <hornschnorter@gmail.com>
Cc: Tom Lane <tgl@sss.pgh.pa.us>, pgsql-hackers@postgreSQL.org
Date: 2018-10-09T23:27:06Z
Lists: pgsql-hackers
Hi,

On 2018-10-09 16:17:44 -0700, Mark Dilger wrote:
> > On Oct 9, 2018, at 12:22 PM, Andres Freund <andres@anarazel.de> wrote:
> > As discussed below (at [1]), I think we should remove $subject.  I plan
> > to do so, unless somebody protests soon-ish.  I thought it'd be better
> > to call attention to this in a new thread, to make sure people had a
> > chance to object.
> 
> I have no objection, but I'm curious, when retiring a datatype and
> associated functions, do the Oids that were assigned to them become
> available for new uses, or do you have to expire them to avoid breaking
> pg_upgrade and such?  Retiring built-in types and functions seems
> rare enough that I've not seen how this is handled before.

I don't really see a need for preserving them. pg_upgrade should fail
because the schema dump won't restore (as that has textual oids). You
could argue that external drivers could have the oids builtin, but I
don't find that convincing, because they'd be in trouble for new types
etc anyway.

Greetings,

Andres Freund


Commits

  1. Clean up/tighten up coercibility checks in opr_sanity regression test.

  2. Remove abstime, reltime, tinterval tables from old regression databases.