From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | pgsql-hackers(at)postgreSQL(dot)org |
Subject: | pg_dump vs casts in extensions |
Date: | 2011-05-16 14:57:36 |
Message-ID: | 15877.1305557856@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
A discussion over in pgsql-admin led me to realize that there's a
corner case bug in pg_dump for casts that belong to extensions.
Namely, pg_dump has a rather ad-hoc rule for deciding whether to
dump a cast, and that rule could result in not dumping a cast belonging
to an extension (if the extension installs its stuff into pg_catalog).
Normally this wouldn't matter since extension members don't get dumped
anyway; but during a binary upgrade, it would result in loss of the
cast.
I think the correct fix is to only apply the namespace heuristic to
casts that aren't members of extensions. If they are, just obey the
"dobj.dump" flag as-is.
There are some other object types like languages that also have
heuristic dump rules, and may need a similar hack.
Barring objections, I'll go make this change.
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Bruce Momjian | 2011-05-16 15:10:03 | pg_upgrade permission check |
Previous Message | Nick Raj | 2011-05-16 14:57:21 | Re: DETOAST Datum |