On Tue, 2008-11-25 at 19:02 +0530, Pavan Deolasee wrote:
> On Tue, Nov 25, 2008 at 6:55 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Huh? The "read only" transaction mode is not hard read-only
> anyway,
> so if that's the only step being taken, it's entirely useless.
>
>
> I think there are explicit checks for some utility statements (like
> VACUUM), but I haven't checked if all necessary code paths are covered
> or not.
The commands that need protecting have been explicitly identified in the
notes and there are 7 files changed that were specifically identified
with protective changes.
You've identified a way of breaking part the first line of defence, but
the command was caught by the second line of defence in the patch.
Problem, yes. Major issue, no. Will fix.
-- Simon Riggs www.2ndQuadrant.comPostgreSQL Training, Services and Support