On Wed, 1 Oct 2003, Tom Lane wrote:
> Stephan Szabo <sszabo@megazone.bigpanda.com> writes:
> > On Tue, 30 Sep 2003, Tom Lane wrote:
> >> I think I can implement it and it will act as stated in my proposal.
> >> Whether people like the proposed behavior is the big question in my
> >> mind.
>
> > I think it's more reasonable than the current behavior or any of
> > the others we've hit along the way, and we have to pretty much choose
> > now if we want to change it for 7.4.
>
> I've committed the attached patch. One thing I wanted to double-check
> with you is that the SELECT FOR UPDATES done in the noaction cases are
> being correctly handled. I think it is correct to do them with the
> current snapshot rather than the start-of-transaction snap; do you
> agree? Also, I did not propagate the crosscheck support into
I think the ones in the main functions need to be current snapshot. I
think the one in ri_Check_Pk_Match doesn't need to be. That's there to see
if this same transaction has inserted a new row with the old value of the
updated/deleted pk row and the serializable snapshot should be fine.
Any conflicting attempts from another transaction should be waiting on our
completion due to the unique index I think.
> heap_mark4update, meaning that these SELECT FOR UPDATEs won't complain
> if they find a row that was inserted later than the start of the
> serializable transaction. I'm not totally sure if they should or not;
> what do you think?
Well, I think that not doing so would only change the error from a
serialization error to a matching row exists error. It might be a bit
surprising if you've just done a select yourself and seen that there were
no matching rows, but I'm not sure that it's a big deal as long as it
errors as appropriate.