> I think Jan muttered something about emitting a warning notice for an
> attempt to store into a table that has an ON SELECT DO
> INSTEAD rule but
> no ON INSERT rule --- which would imply that you'll never be able to
> see the data you're inserting.
>
> This mistake has bitten enough people (including me ;-)) that it seems
> a warning might be a good idea. I'm not sure if I want it to
> be a hard
> error though. Are there any cases where it'd make sense to
> allow this?
Of course the real nice answer would be to create a new pg_class type 'V'
in addition to the existing table types.
Advantage:
1. no table files needed
2. we know it is a view for sure
I for one have tables with "on select do instead" rules that are not views.
Andreas