Re: Distinguish publication exclusions in object addresses - Mailing list pgsql-hackers

From Manuel Reyes Bravo
Subject Re: Distinguish publication exclusions in object addresses
Date
Msg-id CA+bCEdAb_aQPAjoqcpm1PZQdDVnC4biKKNYYD-wBu7JR_z5Mng@mail.gmail.com
Whole thread
In response to Re: Distinguish publication exclusions in object addresses  (vignesh C <vignesh21@gmail.com>)
Responses Re: Distinguish publication exclusions in object addresses
List pgsql-hackers
vignesh C <vignesh21@gmail.com> wrote:
> That seems better, here is an updated v3 version to handle the same.
> Same patch applies on pg19 branch and tests passes in both the
> branches.

I tested v3 on master at bca67e5a33b and on REL_19_STABLE at
1d094904bc7: it applies cleanly, builds without warnings, and make check
passes on both.  With a relation name and a schema name that need
quoting, and a temporary table, the message goes from

    cannot specify relation "public."Part One"" in the publication EXCEPT clause
    cannot specify relation ""My Schema"."we""ird"" in the publication
EXCEPT clause
    cannot specify relation "pg_temp.tmp" in the publication EXCEPT clause

to

    cannot specify relation "public.Part One" in the publication EXCEPT clause
    cannot specify relation "My Schema.we"ird" in the publication EXCEPT clause
    cannot specify relation "pg_temp_0.tmp" in the publication EXCEPT clause

and the FOR TABLE message is unchanged.

v3 no longer has the temporary table test that v2 added, I assume because
the pg_temp_N number depends on the backend.  stats_ext.sql already deals
with that by redacting the number, and the attached top-up does the same,
so the case stays covered:

    NOTICE:  cannot specify relation
"pg_temp_REDACTED.testpub_temptbl" in the publication EXCEPT clause
(This operation is not supported for temporary tables.)

It passes on both branches, and the same block run from three sessions
at once, with temp schemas pg_temp_1, pg_temp_2 and pg_temp_3, prints the
same line in each.

Regards,
Manu

Attachment

pgsql-hackers by date:

Previous
From: Manuel Reyes Bravo
Date:
Subject: pg_get_object_address reports a published relation as non-existent
Next
From: Dilip Kumar
Date:
Subject: Re: Proposal: Conflict log history table for Logical Replication