Re: slru.c race condition

alvherre(at)alvh(dot)no-ip(dot)org (Alvaro Herrera) writes:

> Chris Browne wrote:
>> jnasby(at)pervasive(dot)com ("Jim C. Nasby") writes:
>>
>> > On Tue, Nov 01, 2005 at 11:23:55AM -0300, Alvaro Herrera wrote:
>> >> Tom Lane wrote:
>> >> > "Jim C. Nasby" <jnasby(at)pervasive(dot)com> writes:
>> >> > > AFAIK they're not using subtransactions at all, but I'll check.
>> >> >
>> >> > Well, yeah, they are ... else you'd never have seen this failure.
>> >>
>> >> Maybe it's in plpgsql EXCEPTION clauses.
>> >
>> > Err, I forgot they're using Slony, which is probably using savepoints
>> > and/or exceptions.
>>
>> Slony-I does use exceptions in pretty conventional ways; it does *not*
>> make any use of subtransactions, because it needs to run on PG 7.3 and
>> 7.4 that do not support subtransactions.
>
> Hmm, does it use the BEGIN/EXCEPTION/END construct at all? Because if
> it does, it won't work on 7.4; and if it doesn't, then it isn't using
> savepoints in 8.0 either.

Ah, then I was misreading that.

There are instances of RAISE EXCEPTION, which was what I had in mind,
but not of BEGIN/EXCEPTION/END.

There is some logic in 8.x to *detect* the nesting of transactions,
but that's quite another matter.
--
output = ("cbbrowne" "@" "acm.org")
http://cbbrowne.com/info/multiplexor.html
"If I could find a way to get [Saddam Hussein] out of there, even
putting a contract out on him, if the CIA still did that sort of a
thing, assuming it ever did, I would be for it." -- Richard M. Nixon