On 11/16/2011 01:28 PM, Daniel Farina wrote:
> As it would turn out, a patch for this has already been submitted:
> http://archives.postgresql.org/pgsql-hackers/2011-10/msg00001.php
>
> There was some wrangling on whether it needs to be extended to be
> useful, but for our purposes the formulation already posted already
> captures vital value for us, and in that form appears to be fairly
> uncontentious. I have moved it to the current commitfest, with a
> comment linking to the 'please revive this patch' thread whereby a
> second glance at what to do about this was conducted.
Yeah, that one got a raw deal; it should be in the *current*
CommitFest--you had it in the next one. I'll join the chorus to just
allow people to fire just the pg_cancel_backend pea shooter foot gun
targeted only at their own feet, make most users happy, and punt
anything more complicated off as troublesome relative to its benefit.
That's what Daniel has said, what his co-worker Edward also implemented,
what Noah thought was good enough given other security mechanisms in
place, and what Tom thinks is reasonable too.
This I feel is important, so I'm going to add myself as the next
reviewer and include it in my testing run tomorrow.
--
Greg Smith 2ndQuadrant US greg@2ndQuadrant.com Baltimore, MD
PostgreSQL Training, Services, and 24x7 Support www.2ndQuadrant.us
