Re: [SQL] Calculation dependencies in views

Jan Wieck wrote:
> 
> Rick Delaney wrote:
> 
> > Jan Wieck wrote:
> > >
> > >      Be aware that 6.5.* might be unable to deal with the resulting size of
> > >      the rewrite rule. And I'm not sure that I'll get TOAST ready for 7.0 to
> > >      handle it.
> >
> > I'd have to take a good look at all the expressions again to know if I could do
> > this but I think I probably can.  Only, with what I've started on I'm already
> > getting "rule plan string too big" errors.  (What I've started on is just one
> > view that doesn't have all the expressions in it yet).
> 
>      That's what I meant with "might be unable...". You talked about a 50
>      column view with complicated expressions, and that's surely too large for
>      the current implementation.
> 
> > So I will just get the same thing with nested views, no?  That is how I                          ^^^^^^^^^^
          ^^
 
> > understand the "The Postgres Rule System" section of the manual and Tom Lane's
> > answer to my post.
> 
>      As the author of "The Postgres Rule System" I say yes, except that it
     ^^^^^^^^^
 
>      requires some recursions in the rewriter. Thus, it will take a little

Sorry to be stupid but with my terribly phrased question I can't tell if you mean
"Yes, you are correct that you will still get 'rule plan string too big' errors
with nested views" or "Yes, nested views will work fine as I described despite
your erroneous interpretation of what you read (but you may have to fix the
recursion limit as I already said)".  :-)

I'm 99% sure you mean the latter, now, after re-reading "The Postgres Rule
System" and your first reply to me.  I guess the only reason I'm not 100% sure is
I don't understand where the 'rule plan string too big' limit comes from.  Is it
in the storage of the view/rule (related to the 8K limit on row size)?  I guess
that would make sense.

I suppose if I re-read enough times I'll get it.

Thanks again,

--Rick