On 8 February 2016 at 14:49, Tom Lane <t...@sss.pgh.pa.us> wrote: > Yup. The output column type of the sub-SELECT is determined without > reference to its context, so there's nothing causing the unknown-type > literal to get assigned a definite type.
Mm. I can follow that, although it makes me unhappy that casting the literal to a known type fixes this, it seems unintuitive. > There's been occasional discussion of changing that behavior, but it's > not real clear that it wouldn't create as many problems as it solves. A more simple solution (to my problem, at least!) might be to stop COALESCE trying to coerce NULLs into a type at all. I don't see how that could ever cause any problems, since NULL is only ever discarded in this context. I would understand it would be difficult if the coercion is taking place at a higher level, but I don't see how that can be the case, because the type it tries to coerce the NULL into is defined by the second argument (which must be COALESCE-specific behaviour, I would think). Geoff -- Sent via pgsql-general mailing list (pgsql-general@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
