Re: tab complete for COPY populated materialized view TO

On Thursday, April 10, 2025, Kirill Reshke <reshkekirill(at)gmail(dot)com> wrote:

> On Thu, 10 Apr 2025 at 20:07, Fujii Masao <masao(dot)fujii(at)oss(dot)nttdata(dot)com>
> wrote:
> >
> >
> >
> > On 2025/04/09 19:24, Kirill Reshke wrote:
> > > On Wed, 9 Apr 2025 at 14:45, Fujii Masao <masao(dot)fujii(at)oss(dot)nttdata(dot)com>
> wrote:
> > >>
> > >>
> > >>
> > >> On 2025/04/09 18:25, Kirill Reshke wrote:
> > >>> On Wed, 9 Apr 2025 at 13:23, jian he <jian(dot)universality(at)gmail(dot)com>
> wrote:
> > >>>>
> > >>>> hi.
> > >>>>
> > >>>> we allow the "COPY table TO" command to copy rows from materialized
> > >>>> views in [1].
> > >>>> The attached patch is to add a tab complete for it.
> > >>>>
> > >>>> [1] https://git.postgresql.org/cgit/postgresql.git/commit/?id=
> 534874fac0b34535c9a5ab9257d6574f78423578
> > >>>
> > >>> Hi!
> > >>> Patch works good for me, but I noticed that psql COPY <tab> suggests
> > >>> partitioned relation both with and without this patch. Maybe that's
> > >>> not a big problem, if [0] will be pushed.
> > >>
> > >> Is the partitioned table currently tab-completed for the COPY FROM
> case?
> > >
> > > If I'm not mistaken, yes. I double checked.
> > >
> > >> INSTEAD OF INSERT triggers - though maybe that's overkill?
> > >
> > > That's wild to me, psql tab completions feature designed to support
> > > postgresql not fully, but in frequent cases. So maybe we should keep
> > > it stupud.
> >
> > I agree that it's reasonable to exclude such rarely used objects from
> > tab-completion. How about including just tables, partitioned tables,
> > foreign tables, and materialized views?
> > I've attached a patch for that.
> >
> > Regards,
>
> Patch is ok. However...

I concur with the premise of the patch. Tab-complete is going to happen
before we know whether to/from is specified so the syntax limits our smarts
here.

> > If we aim to support tab-completion for all valid targets of both COPY TO
> and COPY FROM, shouldn't foreign tables also be included?
>
> Ah.. Sorry I missed this part of your message initially. No, foreign
> tables are not supported:

They are supported for the From variant; valid completions need only
satisfy one of to/from, not both.

>
> What's funny is that copying foreign tables using MV works fine
>
> ```
> reshke=# create materialized view mv as table ft;
> SELECT 1
> reshke=# copy mv to stdout;
> 228
> ```
>

I don’t get why this is “funny” or otherwise surprising.

David J.