Horton

Re: POC PATCH: copy from ... exceptions to: (was Re: VLDB Features)

vignesh C <vignesh21@gmail.com>

View thread
From: vignesh C <vignesh21@gmail.com>
To: jian he <jian.universality@gmail.com>
Cc: Masahiko Sawada <sawada.mshk@gmail.com>, torikoshia <torikoshia@oss.nttdata.com>, Alena Rybakina <lena.ribackina@yandex.ru>, Damir Belyalov <dam.bel07@gmail.com>, zhihuifan1213@163.com, Tom Lane <tgl@sss.pgh.pa.us>, Daniel Gustafsson <daniel@yesql.se>, PostgreSQL Hackers <pgsql-hackers@postgresql.org>, Andres Freund <andres@anarazel.de>, anisimow.d@gmail.com, HukuToc@gmail.com, pgsql-hackers@lists.postgresql.org, Andrei Lepikhov <a.lepikhov@postgrespro.ru>
Date: 2024-01-04T16:05:20Z
Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. doc: Fix COPY ON_ERROR option syntax synopsis.

  2. Disallow specifying ON_ERROR option without value.

  3. Rename COPY option from SAVE_ERROR_TO to ON_ERROR

  4. Fix spelling in notice

  5. Add new COPY option SAVE_ERROR_TO

On Thu, 28 Dec 2023 at 09:27, jian he <jian.universality@gmail.com> wrote:
>
> On Wed, Dec 20, 2023 at 8:27 PM Masahiko Sawada <sawada.mshk@gmail.com> wrote:
> >
> >
> > Why do we need to use SPI? I think we can form heap tuples and insert
> > them to the error table. Creating the error table also doesn't need to
> > use SPI.
> >
> Thanks for pointing it out. I figured out how to form heap tuples and
> insert them to the error table.
> but I don't know how to create the error table without using SPI.
> Please pointer it out.
>
> > >
> > > copy_errors one per schema.
> > > foo.copy_errors will be owned by the schema: foo owner.
> >
> > It seems that the error table is created when the SAVE_ERROR is used
> > for the first time. It probably blocks concurrent COPY FROM commands
> > with SAVE_ERROR option to different tables if the error table is not
> > created yet.
> >
> I don't know how to solve this problem.... Maybe we can document this.
> but it will block the COPY FROM immediately.
>
> > >
> > > if you can insert to a table in that specific schema let's say foo,
> > > then you will get privilege to INSERT/DELETE/SELECT
> > > to foo.copy_errors.
> > > If you are not a superuser, you are only allowed to do
> > > INSERT/DELETE/SELECT on foo.copy_errors rows where USERID =
> > > current_user::regrole::oid.
> > > This is done via row level security.
> >
> > I don't think it works. If the user is dropped, the user's oid could
> > be reused for a different user.
> >
>
> You are right.
> so I changed, now the schema owner will be the error table owner.
> every error table tuple inserts,
> I switch to schema owner, do the insert, then switch back to the
> COPY_FROM operation user.
> now everyone (except superuser) will need explicit grant to access the
> error table.

There are some compilation issues reported at [1] for the patch:
[04:04:26.288] copyfromparse.c: In function ‘NextCopyFrom’:
[04:04:26.288] copyfromparse.c:1126:25: error: ‘copy_errors_tupDesc’
may be used uninitialized in this function
[-Werror=maybe-uninitialized]
[04:04:26.288] 1126 | copy_errors_tup = heap_form_tuple(copy_errors_tupDesc,
[04:04:26.288] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[04:04:26.288] 1127 | t_values,
[04:04:26.288] | ~~~~~~~~~
[04:04:26.288] 1128 | t_isnull);
[04:04:26.288] | ~~~~~~~~~
[04:04:26.288] copyfromparse.c:1160:4: error: ‘copy_errorsrel’ may be
used uninitialized in this function [-Werror=maybe-uninitialized]
[04:04:26.288] 1160 | table_close(copy_errorsrel, RowExclusiveLock);
[04:04:26.288] | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

[1] - https://cirrus-ci.com/task/4785221183209472

Regards,
Vignesh