PG Bug reporting form <noreply@postgresql.org> writes:
> create operator qwe.= (leftarg = char, rightarg = text, function =
> qwe.chartexteq, commutator = operator(qwe.=), hashes, merges);
> set search_path = qwe;
> explain (costs off, verbose on) select i from generate_series(1, 10) i where
> i::char in (2::text);
> I expected, that IN list translates to pg_catalog.=
Why would you expect that? It'd make it impossible to use IN
with user-defined data types. In this case, you made an operator
that is a closer match to the given datatypes (ie, "char = text")
than the native "text = text" operator, so it used that one.
I've not checked the code, but my recollection is that X IN (Y) just
resolves to the same equality operator you'd get by writing X = Y.
There's been some discussion about allowing a schema qualifier to
be included in the syntax, but nothing's been done about that.
regards, tom lane