Vik Fearing <vik@postgresfriends.org> writes:
> On 10/2/23 21:25, Tom Lane wrote:
>> Sorry for not weighing in on this before, but ... is this a feature
>> we want at all?
> For standards conformance, I vote yes.
Only if we can actually implement it in a defensible way, which this
patch is far short of accomplishing.
>> We are very clear in the existing docs that CHECK
>> conditions must be immutable [1],
> That is what the *user* documentation says, but we both know it isn't true.
> Here is a short conversation you and I had about five years ago where
> you defended the non-immutability of CHECK constraints:
> https://www.postgresql.org/message-id/flat/12539.1544107316%40sss.pgh.pa.us
What I intended to defend was not *checking* immutability strictly.
Our CHECK constraint implementation is based very much on the assumption
that the constraints are immutable, and nobody has proposed that we
try to remove that assumption AFAIR. So I think the docs are fine
as-is; anybody who wants to get into monotonically-weakening constraints
is probably smart enough to work out for themselves whether it will
fly or not.
So my problem with this patch is that it does nothing about that
assumption, and yet the feature it adds seems useless without
weakening the assumption. So what weaker assumption could we
make, and how would we modify the when-to-check rules to match
that, and what would it cost us in performance? Without good
answers to those questions, this patch is just a facade.
> I disagree with this. The whole point of deferring constraints is to be
> able to do some cleanup before the consistency is checked.
What cleanup would you need that couldn't be performed beforehand
(e.g. in a BEFORE INSERT/UPDATE trigger)? All the practical
examples that occur to me involve cross-row conditions, which
CHECK is unsuitable to enforce --- at least, without doing a
thorough implementation rethink.
I continue to assert that basing this feature on the current
CHECK implementation will produce nothing but a toy feature,
that's not only of little practical use but will be an active
foot-gun for people who expect it to do more than it can.
regards, tom lane