William Bug wrote:
>
>
> I'm not certain I understand what you mean here? Are you recommending
> all application layer interaction with tables using INHERIT should be
> done via a VIEW intermediary? If so, wouldn't the VIEW (built from a
> "SELECT ... ONLY...") then be as dependent on the fixed structure
> determined by the INHERITs relationship, as much as the application
> code would be?
>
Well, what I'm concerned about is this: you have an inheritance
hierarchy in PG, and some application has a "SELECT ... ONLY" in it. If
you want to change the inheritance hierarchy in PG around, you may not
be able to make it "look like" the old hierarchy to the application with
views.
If you use a view in between, maybe the view does the "SELECT ... ONLY".
That way, if you change the PG inheritance hierarchy, you can just
change the view without changing the application. The application would
never use "SELECT ... ONLY" so you would always have a way out if you
need it.
My basic philosophy here is that as long as you have a way out, it's not
wrong.
Regards,
Jeff Davis
---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match
