On Fri, Jan 18, 2019 at 4:50 AM Vik Fearing <vik.fearing@2ndquadrant.com> wrote:
> Isn't that what happens already? PrepareTransaction() calls
> PreCommit_on_commit_actions() from what I can tell.
Huh. Well, in that case, I'm not sure I understand we really need to
do beyond removing the error checks for the case where all tables are
on-commit-drop.
It could be useful to do something about the issue with pg_temp
creation that Tom linked to in the other thread. But even if you
didn't do that, it'd be pretty easy to work around this in application
code -- just issue a dummy CREATE TEMP TABLE .. ON COMMIT DROP
statement the first time you use a connection, so that the temp schema
definitely exists. So I'm not sure I'd view that as a blocker for
this patch, even though it's kind of a sucky limitation.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company