On Wed, Aug 21, 2013 at 10:13:15AM -0400, Tom Lane wrote:
> Alexander Korotkov <aekorot...@gmail.com> writes:
> > It seems reasonable to me to cast enum to oid. However, creating casts
> > without function isn't allowed for enums.
>
> > test=# create cast (source as oid) without function;
> > ERROR: enum data types are not binary-compatible
>
> The reason for that is you'd get randomly different results on another
> installation. In this particular application, I think David doesn't
> really care about what values he gets as long as they're distinct,
> so this might be an OK workaround for him. But that's the reasoning
> for the general prohibition.
While a WITHOUT FUNCTION cast does *guarantee* that flaw, working around the
restriction with a cast function is all too likely to create the same flaw.
Here's the comment about the restriction:
* Theoretically you could build a user-defined base type that
is
* binary-compatible with a composite, enum, or array type.
But we
* disallow that too, as in practice such a cast is surely a
mistake.
* You can always work around that by writing a cast function.
That's reasonable enough, but we could reduce this to a WARNING. Alexander
shows a credible use case. A superuser can easily introduce breakage through
careless addition of WITHOUT FUNCTION casts. Permitting borderline cases
seems more consistent with the level of user care already expected in this
vicinity.
--
Noah Misch
EnterpriseDB http://www.enterprisedb.com
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
