neil(at)postgrescompare(dot)com writes:
> This is probably not a bug
Nope, it isn't.
> If it's not a bug should postgres support
> ALTER MATERIALIZED VIEW ALTER COLUMN column SET NOT NULL syntax?
Doesn't seem like a great idea to me. What's supposed to happen
if a view update would result in a violation of the constraint?
It's way too late to undo whatever change of the underlying data
caused that, so it seems like the only possible answer is for
REFRESHes to fail, resulting in the view getting more and more
out of date until and unless something manual is done to resolve
the problem. I can't see that that's a useful behavior to have.
regards, tom lane