On 2010-08-25 12:44 PM +0300, Heikki Linnakangas wrote:
> On 25/08/10 12:41, Andres Freund wrote:
>> But randomly loosing tuples will make much more people unhappy. At a
>> much more problematic point of time (in production).
>
> Hmm, how would you lose tuples?
I think what Andres means is: T1 starts a MERGE. INSERT fails because
the tuple already exists, but then another transaction, T2, DELETEs that
tuple. T1 tries to UPDATE it, but fails because it doesn't exist
anymore. Not T1 should go back and INSERT the tuple, but that isn't
what happens with this patch, is it?
Regards,
Marko Tiikkaja