Re: Protection lost in expression eval changeover

On 2017-03-28 14:43:38 -0400, Tom Lane wrote:
> Andres Freund <andres(at)anarazel(dot)de> writes:
> > On 2017-03-28 13:52:50 -0400, Tom Lane wrote:
> >> So it seems like we are missing some needed protection. I'm inclined
> >> to think that it'd be all right to just throw an error immediately in
> >> CheckVarSlotCompatibility if the target column is dropped.
>
> > Hm - so far we've pretty widely only set columns to NULL in that
> > case. You don't see concerns with triggering errors in cases we
> > previously hadn't?
>
> Well, in principle these errors ought to be unreachable at all; they're
> only there to backstop any possible failure to notice stale plans.

Not entirely - e.g. I think some of that's reachable when composite
types are involved. But it's probably ok to error out anyway.

> I don't see a strong reason why we need to allow a dropped column to go
> to null while we throw an immediate error for a change in column type.
> (If there is some reason, hopefully beta testing will find it.)

Ok. You're doing that?

Greetings,

Andres Freund