On Nov 17, 2009, at 9:48 AM, Justin Mason wrote:
On Tue, Nov 17, 2009 at 13:36, Mark Martinec <Mark.Martinec
[email protected]> wrote:
About the 6203, I have some stats collected. Also the AWL race
condition:
https://issues.apache.org/SpamAssassin/show_bug.cgi?id=4642#c1
is worrying me. I'm using a workaround (attempt INSERT first, and
if that fails do the UPDATE), but the change is nontrivial, and
depends
on a unique key constraint failing to work properly. If you like,
I can submit it for consideration - can be reverted later if not
approved.
An alternative is to use ON DUPLICATE KEY UPDATE, but that seems not
to be liked according to Bug 5998.
If you can provoke Michael into a response ;) and get agreement,
then whatever
you guys decide is good. However I don't think it is safe to delay
beta release
for this, as it's a pretty thorny problem AFAICT.
Ok, consider me provoked :)
Commented on bug 5998, my objection wasn't the ON DUPLICATE KEY UPDATE
call, it was on the fact that it would make master-master setups work.
As to the concurrency issues, I learned just enough plpgsql to come up
with those functions and have probably forgotten it, someone with
better PostgreSQL skills needs to pick up the ball and figure things
out.
Michael
