On Sat, Jan 05, 2019 at 09:00:37AM +0900, Michael Paquier wrote:
> I think I can get that worked out with a back-patchable approach,
> still it looks a bit sensitive because of the creation-pending logic
> which relies on the assertion at the top of InitTempTableNamespace,
> which is a case we may want to handle with an extra flag for the
> caller of InitTempTableNamespace(), aka "fail if myTempNamespace is
> valid instead of just returning back".
Looking at this stuff, I have been able to come up with the attached,
which introduces a new flag mode for MyXactFlags, which gets set in a
couple of code paths so as PREPARE TRANSACTIOn complains:
- When attempting a LOCK on a temporary relation.
- When dropping objects part of a temporary namespace.
- When trying to access the temporary schema of a session, where it
may be initialized (or not if already present).
I have added test cases for things I could come up with, particularly
the LOCK and DROP cases which I have thought about after hacking the
code. I think that something among those lines should be
back-patched.
Thoughts?
--
Michael