Stephen Frost <[EMAIL PROTECTED]> writes:
> In doing some test on 8.3RC2, I was dismayed to discover that the
> pl/pgsql plan invalidation logic added doesn't consider changing the
> search_path to invalidate a plan.
We never considered it so before, either. The plancache code goes out
of its way to maintain the same path that was used initially, and
I think that's what it should do: a cache module should avoid letting
the semantics of what it's cached change without the caller's knowledge.
If we were to change this, we'd probably have to think in terms of
making the active search_path be part of the lookup key for cached plans.
> Would it be possible to have this case handled?
It's far too late to reconsider this point for 8.3. If you want to
bring it up for 8.4, we could think about what the behavioral and
performance implications would really be. In the meantime, the answer
is the same as it's always been: if that's what you want, use EXECUTE.
regards, tom lane
---------------------------(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
