Re: Another issue in default-values patch: defaults expanded too soon

Josh Berkus <josh@agliodbs.com> writes:
> Huh?  Shouldn't changing a function which a view depends on require a 
> rebuild of the view?

There is no such concept as "a rebuild of the view".

> That is, I think we should treat changing the defaults the same as we 
> would changing the number and type of parameters; it kicks off a 
> dependency check and requires a CASCADE.

Dream on ... there is no such facility in Postgres and we are not going
to build one in the 8.4 timeframe.
        regards, tom lane