Re: Bug in pg_dump/restore -o - Mailing list pgsql-hackers

From Tom Lane
Subject Re: Bug in pg_dump/restore -o
Date
Msg-id 4002.1011327869@sss.pgh.pa.us
Whole thread Raw
In response to Re: Bug in pg_dump/restore -o  (Philip Warner <pjw@rhyme.com.au>)
List pgsql-hackers
Philip Warner <pjw@rhyme.com.au> writes:
> A sigle TOC entry, with both CREATE and COPY run at the start no matter
> what kind of dump/restore, seems like the solution. I'll look into it.

Okay.  Keep in mind that given where we are in the release cycle, any
patch applied now has got to be minimal and easily checked.  If what
you are thinking of involves any extensive changes, it'd probably be
better to let the problem go till 7.3, rather than risk introducing
worse breakage in 7.2.  Especially since this problem isn't that major
in the big scheme of things (if it were, it'd have been found sooner).

One stopgap answer Bruce and I discussed on the phone was to temporarily
add a couple lines of code to pg_dump to reject the combination of -o
with -Fc or -Ft.  That would prevent people from shooting themselves in
the foot in 7.2 until a proper solution can be devised for 7.3.
        regards, tom lane


pgsql-hackers by date:

Previous
From: Bruce Momjian
Date:
Subject: Re: Bug in pg_dump/restore -o
Next
From: Bruce Momjian
Date:
Subject: Re: tuptoaster.c must *not* use SnapshotAny