Re: Partitioning with temp tables is broken

On 2018/06/20 10:54, Michael Paquier wrote:
> On Tue, Jun 19, 2018 at 04:26:56PM +0530, Ashutosh Bapat wrote:
>> On Tue, Jun 19, 2018 at 4:22 PM, Michael Paquier <michael@paquier.xyz> wrote:
>>> I was under the impression that this was implied in the precious
>>> phrasing but you guys visibly don't match with my impression.  So I
>>> would suggest this paragraph at the end:
>>> "Mixing temporary and permanent relations in the same partition tree is
>>> not allowed.  Hence, if the partitioned table is permanent, so must be
>>> its partitions at all levels and likewise if the partitioned table is
>>
>> You don't need to mention "at all levels", its implied recursively.
> 
> Okay, done on master and REL_10_STABLE.  I have adapted the tests and
> the code on v10 where default partitions do not apply.  I have also
> removed the test case for partition pruning in REL_10_STABLE as those
> have been mainly added by 8d4e70a6, master of course keeps it.

Thank you, especially for putting in the extra work for back-patching.  I
shouldn't have used default partition syntax in tests, sorry.

> I have included Ashutosh's last suggestions and finished with the
> following phrasing:

I liked both of Ashutosh's suggestions, which I see you incorporated into
the commit.

> "Mixing temporary and permanent relations in the same partition tree is
> not allowed.  Hence, if the partitioned table is permanent, so must be
> its partitions and likewise if the partitioned table is temporary.  When
> using temporary relations, all members of the partition tree have to be
> from the same session."

Just a minor nit in the last sentence:

"have to be from" -> "must be from / must belong to"

> I am not sure why this set of emails does not actually appear on
> UI interface for archives of pgsql-hackers...  All hackers are receiving
> that, right?

I evidently got your email just fine. :)

Thanks,
Amit