Re: Declarative partitioning - another take

On Fri, Apr 28, 2017 at 06:29:48PM +0900, Amit Langote wrote:
> On 2017/04/28 7:36, David Fetter wrote:
> > On Thu, Apr 27, 2017 at 10:30:54AM +0900, Amit Langote wrote:
> >> On 2017/04/27 1:52, Robert Haas wrote:
> >>> On Tue, Apr 25, 2017 at 10:34 PM, Amit Langote
> >>> <Langote_Amit_f8(at)lab(dot)ntt(dot)co(dot)jp> wrote:
> >>>> FWIW, I too prefer the latter, that is, fire only the parent's triggers.
> >>>> In that case, applying only the patch 0001 will do.
> >>>
> >>> Do we need to update the documentation?
> >>
> >> Yes, I think we should. How about as in the attached?
> >>
> >> By the way, code changes I made in the attached are such that a subsequent
> >> patch could implement firing statement-level triggers of all the tables in
> >> a partition hierarchy, which it seems we don't want to do. Should then
> >> the code be changed to not create ResultRelInfos of all the tables but
> >> only the root table (the one mentioned in the command)? You will see that
> >> the patch adds fields named es_nonleaf_result_relations and
> >> es_num_nonleaf_result_relations, whereas just es_root_result_relation
> >> would perhaps do, for example.
> >
> > Did I notice correctly that there's no way to handle transition tables
> > for partitions in AFTER STATEMENT triggers?
>
> Did you mean to ask about AFTER STATEMENT triggers defined on
> "partitioned" tables? Specifying transition table for them is disallowed
> at all.
>
> ERROR: "p" is a partitioned table
> DETAIL: Triggers on partitioned tables cannot have transition tables.

OK, I suppose. It wasn't clear from the documentation.

> Triggers created on (leaf) partitions *do* allow specifying transition table.

That includes the upcoming "default" tables, I presume.

> Or are you asking something else altogether?

I was just fuzzy on the interactions among these features.

> > If not, I'm not suggesting that this be added at this late date, but
> > we might want to document that.
>
> I don't see mentioned in the documentation that such triggers cannot be
> defined on partitioned tables. Is that what you are saying should be
> documented?

Yes, but I bias toward documenting a lot, and this restriction could
go away in some future version, which would make things more confusing
in the long run. I'm picturing a conversation in 2020 that goes
something like this:

"On 10, you could have AFTER STATEMENT triggers on tables, foreigh
tables, and leaf partition tables which referenced transition tables,
but not on DEFAULT partitions. On 11, you could on DEFAULT partition
tables. From 12 onward, you can have transition tables on any
relation."

Kevin? Thomas?

Best,
David.
--
David Fetter <david(at)fetter(dot)org> http://fetter.org/
Phone: +1 415 235 3778 AIM: dfetter666 Yahoo!: dfetter
Skype: davidfetter XMPP: david(dot)fetter(at)gmail(dot)com

Remember to vote!
Consider donating to Postgres: http://www.postgresql.org/about/donate