Richard Huxton wrote:
Heikki Linnakangas wrote:
The problem is that the new tuple version is checked only against the
condition in the update rule, id=OLD.id, but not the condition in the
original update-claus, dt='a'.
Yeah, that's confusing :(.
Bit more than just normal rule confusion I'd say. Try the following two
statements in parallel (assuming you've just run the previous):
UPDATE test SET dt='c';
UPDATE test SET dt='x' FROM test t2 WHERE test.id=t2.id AND t2.dt='b';
This isn't a problem with the view mechanism - it's a problem with
re-checking clauses involving subqueries or joins I'd guess.
I'm trying to decide if it's unexpected or just plain wrong, and I think
I'd have to argue wrong.
Or perhaps I'd not argue that :-/
This is really about MVCC in read committed mode, and the "just right
for simpler cases":
Clearly there needs to be a change to the sentence: "Because of the
above rule, it is possible for an updating command to see an
inconsistent snapshot: it can see the effects of concurrent updating
commands that affected the same rows it is trying to update"
Not true if there's a subquery/join involved.
---------------------------(end of broadcast)---------------------------
TIP 3: Have you checked our extensive FAQ?