MIME-Version: 1.0
References: 
 <CAA5RZ0s+kZZRMSF4HW7tZ9W2jS1o4B+Fg8dr5a-T6mANX+mdQA@mail.gmail.com>
 <CAD21AoDZG6CUE9by0T2N33Zedw4iH0JF=9YoCuniXcugO+ixog@mail.gmail.com>
 <CAJDiXgjQEdssGEQ97UZEKBK5Qc5=urqi1qVLdsJwHbQtsBPH1g@mail.gmail.com>
 <CAA5RZ0tATt=YNy9suA3svMAjw3oBDzxcw8+gMMVHbYG9NBQtRA@mail.gmail.com>
 <CAJDiXgibFmjCvpVojOGCU4DfaqJ3pE+p2BTn_spZNVKrbcGjLQ@mail.gmail.com>
 <CAA5RZ0tExiffcu7qvrUbpq_qqz=zCD2aJ5_Qigo6eP2kgTx3eQ@mail.gmail.com>
 <adWcVrX3jrHPoCmD@paquier.xyz>
In-Reply-To: <adWcVrX3jrHPoCmD@paquier.xyz>
From: Sami Imseih <samimseih@gmail.com>
Date: Wed, 8 Apr 2026 09:52:39 -0500
Message-ID: 
 <CAA5RZ0tH1Wb+5b+WW+6Cs9jFRZ215n5g=ZsB9KL7zmp5gBi+kQ@mail.gmail.com>
Subject: Re: test_autovacuum/001_parallel_autovacuum is broken
To: Michael Paquier <michael@paquier.xyz>,
 Masahiko Sawada <sawada.mshk@gmail.com>,
	Daniil Davydov <3danissimo@gmail.com>
Cc: pgsql-hackers <pgsql-hackers@postgresql.org>
Content-Type: text/plain; charset="UTF-8"
Archived-At: 
 <https://www.postgresql.org/message-id/CAA5RZ0tH1Wb%2B5b%2BWW%2B6Cs9jFRZ215n5g%3DZsB9KL7zmp5gBi%2BkQ%40mail.gmail.com>
Precedence: bulk

> On Tue, Apr 07, 2026 at 01:02:49PM -0500, Sami Imseih wrote:
> > Perhaps, but I don't see it being unreasonable for injection points.
> >
> > I guess we can also think about expanding InjectionPointCondition to
> > handle other types of conditions, maybe OID??, to filter when running
> > the point.
>
> Yeah, InjectionPointConditionType was designed with these types of
> extensions in mind in terms of conditions you want to assign to a
> point name when attaching it, with the check happening in the callback
> attached when it is run.  It should not be complicated to extend
> injection_points_attach(), just pass down a string that it then
> translated to an OID, or you could use a different grammar as well.
> One thing that I'd be careful about is to handle that with one
> argument in the SQL attach function, with the condition data given as
> an input string.  One grammar that Alexander K. designed at some point
> for some of the facilities he had proposed was a JSON input, but
> that's an implementation artifact.
>
> Doing something as a separate module/library would be also fine.  We
> do that for the AIO tests.

I think we can enhance these tests using a separate module as Daniil is
suggesting, but we should probably get 0001 committed first and then
have a quick follow-up. What do you think?

--
Sami