Abhijit Menon-Sen wrote:
> At 2016-04-05 12:33:56 +0530, ams@2ndQuadrant.com wrote:
> >
> > Álvaro: I did document and test the extra types you added, but now
> > that I think about it a bit more, it's hard to argue that it's useful
> > to have a table, for example, depend on an extension. There's really
> > nothing about a table that "doesn't work without" an extension.
>
> I think it makes sense to implement this for triggers and functions. It
> may also be useful for indexes and materialised views, which can refer
> to functions in an extension (and in future, sequences as well).
>
> It's certainly good to know the grammar would work if we wanted to add
> other object types in future, but I think we should leave it at that.
Yes, agreed. What I implemented weren't cases that were supposed to be
useful to users, only those for which I thought it was good to test
bison with. Sorry I wasn't clear about this. Feel free the strip out
(some of?) them, if they aren't useful.
--
Álvaro Herrera http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services