Thomas Lockhart <lockhart@alumni.caltech.edu> writes:
> btw, even gram.y does touch some of the heap cache (for pg_type) to look
> for type existance; don't know if that will be a problem but maybe that
> needs to be rethought also...
We'd need to postpone that processing till parse analysis, else we still
have the underlying problem. Fortunately we are not parsing C ;-) so
it seems to me it shouldn't be necessary to do any table lookups during
initial parsing...
I assume you are looking at the 'setof' processing? Offhand it seems to
me that this code is broken anyway: use of a relation type should refer
to the tuple type, but should *not* imply SETOF, at least IMHO.
regards, tom lane