Re: Incremental View Maintenance, take 2 (design considerations)

On Wed, 1 Jul 2026 00:04:01 +0900
Yugo Nagata <nagata(at)sraoss(dot)co(dot)jp> wrote:

> On Fri, 29 May 2026 23:14:17 +0900
> Yugo Nagata <nagata(at)sraoss(dot)co(dot)jp> wrote:
>
> > There are still various design trade-offs to consider, so comments and discussion
> > on the overall direction would be greatly appreciated.
>
> While I'm still working on those design changes and scope reduction,
> I've attached an updated patch to fix the broken test in the previous
> patch set.
>
> I'll send another patch set reflecting those changes once it's ready.

I've attached a reduced patch set excluding DISTINCT and aggregate support to
reduce the review scope.　I'm still working on the planned design changes, and
I'll send another patch set reflecting those changes once they're ready.

Regards,
Yugo Nagata

>
> Regards,
> Yugo Nagata
>
> > On Sun, 22 Feb 2026 23:41:17 +0900
> > Yugo Nagata <nagata(at)sraoss(dot)co(dot)jp> wrote:
> >
> > > On Mon, 16 Feb 2026 21:43:09 +0000
> > > Alexandre Felipe <o(dot)alexandre(dot)felipe(at)gmail(dot)com> wrote:
> > >
> > > Thank you for looking over and updating the patches!
> > > I’m planning an overhaul of the patch set, but I really appreciate your time
> > > and effort in getting them up to date.
> > >
> > > > Sorry,
> > > > the previous line missed the removal of the function declared but not used.
> > > >
> > > > Regards,
> > > > Alexandre
> > > >
> > > > On Mon, Feb 16, 2026 at 4:07 PM Alexandre Felipe <
> > > > o(dot)alexandre(dot)felipe(at)gmail(dot)com> wrote:
> > > >
> > > > > There was a warning on my initial rebase, so I fixed that.
> > > > >
> > > > > I also changed the bitmap set to a list, I don't think we need O(1) lookup
> > > > > here as suggested by Zhang [1] on patch 6.
> > > > >
> > > > > Yugo,
> > > > > I think there is an issue in
> > > > > src/backend/commands/matview.c, IVM_immediate_maintenance, line 1688, when
> > > > > apply_delta fails, and PG_RE_THROW is called, wouldn't we have to cleanup?
> > > > > As in line 1699 onwards?
> > >
> > > Do you mean calling clean_up_IVM_hash_entry() as part of the cleanup?
> > > I would need to look into this more carefully, but my understanding is that i
> > > might be handled by AtAbort_IVM() in that situation.
> > >
> > > Regards,
> > > Yugo Nagata
> > >
> > > > >
> > > > >
> > > > >
> > > > > Regards,
> > > > > Alexandre
> > > > >
> > > > > On Thu, Feb 12, 2026 at 6:08 PM Alexandre Felipe <
> > > > > o(dot)alexandre(dot)felipe(at)gmail(dot)com> wrote:
> > > > >
> > > > >> Sorry for creating a new thread for this.
> > > > >> I don't have the original email.
> > > > >>
> > > > >> This is my attempt on rebasing
> > > > >> https://commitfest.postgresql.org/patch/4337/
> > > > >>
> > > > >> Regards,
> > > > >> Alexandre
> > > > >>
> > > > >
> > >
> > >
> > > --
> > > Yugo Nagata <nagata(at)sraoss(dot)co(dot)jp>
> > >
> > >
> >
> >
> > --
> > Yugo Nagata <nagata(at)sraoss(dot)co(dot)jp>
>
>
> --
> Yugo Nagata <nagata(at)sraoss(dot)co(dot)jp>

--
Yugo Nagata <nagata(at)sraoss(dot)co(dot)jp>