Re: [HACKERS] DROP TABLE inside a transaction block

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: "Hiroshi Inoue" <Inoue(at)tpf(dot)co(dot)jp>
Cc: pgsql-hackers(at)postgreSQL(dot)org
Subject: Re: [HACKERS] DROP TABLE inside a transaction block
Date: 2000-03-08 16:15:56
Message-ID: 25458.952532156@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

"Hiroshi Inoue" <Inoue(at)tpf(dot)co(dot)jp> writes:
> Is it necessary to get the relation path name from the relation name/oid etc
> each time ?
> Is it bad to keep the relation path name in pg_class(or another relation) ?

Hmm, we could maybe do that for user relations, but it obviously would
not work for pg_class itself. I'm a little worried about trying to do
it for the other critical system relations, too. We'd want to keep the
relation's pathname in its relcache entry, so any system relation that
is read while setting up a relcache entry has to have a fixed path that
can be determined without a relcache entry.

Perhaps it would be good enough to say that all system relations live in
the database's primary directory, and only user relations have pathnames
specified in their pg_class entries. Renaming a system table would be
a Really Bad Idea anyway ;-)

regards, tom lane

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Kyle Bateman 2000-03-08 16:22:17 Re: Casts in 7.0 vs 6.5 (was Re: [SQL] 7.0beta bug (or feature)?)
Previous Message Michael Meskes 2000-03-08 15:30:10 Re: [HACKERS] library policy question