Horton

Re: DROP DATABASE is interruptible

Andres Freund <andres@anarazel.de>

View thread View on postgr.es
From: Andres Freund <andres@anarazel.de>
To: Peter Eisentraut <peter@eisentraut.org>
Cc: Daniel Gustafsson <daniel@yesql.se>, Thomas Munro <thomas.munro@gmail.com>, PostgreSQL Hackers <pgsql-hackers@postgresql.org>, Evgeny Morozov <postgresql3@realityexists.net>, Tom Lane <tgl@sss.pgh.pa.us>
Date: 2023-09-25T18:52:23Z
Lists: pgsql-hackers 
Hi,

On 2023-09-25 01:48:31 +0100, Peter Eisentraut wrote:
> I noticed that this patch set introduced this pg_dump test:
> 
> On 12.07.23 03:59, Andres Freund wrote:
> > +	'CREATE DATABASE invalid...' => {
> > +		create_order => 1,
> > +		create_sql => q(CREATE DATABASE invalid; UPDATE pg_database SET datconnlimit = -2 WHERE datname = 'invalid'),
> > +		regexp => qr/^CREATE DATABASE invalid/m,
> > +		not_like => {
> > +			pg_dumpall_dbprivs => 1,
> > +		},
> > +	},
> 
> But the key "not_like" isn't used for anything by that test suite. Maybe
> "unlike" was meant?

It's not clear to me either. Invalid databases shouldn't *ever* be dumped, so
explicitly listing pg_dumpall_dbprivs is odd.

TBH, I find this testsuite the most opaque in postgres...


> But even then it would be useless because the "like" key is empty, so there
> is nothing that "unlike" can subtract from.  Was there something expected
> from the mention of "pg_dumpall_dbprivs"?

Not that I can figure out...


> Perhaps it would be better to write out
> 
>     like => {},
> 
> explicitly, with a comment, like some other tests are doing.

Yea, that looks like the right direction.

I'll go and backpatch the adjustment.

Greetings,

Andres Freund

Commits

  1. pg_dump: tests: Correct test condition for invalid databases

  2. Handle DROP DATABASE getting interrupted