Re: Adding a test for speculative insert abort case

Hi,

On 2019-05-01 11:41:48 -0700, Andres Freund wrote:
> I'm imagining something like
>
> if (pg_try_advisory_xact_lock(1))
> pg_advisory_xact_lock(2);
> else
> pg_advisory_xact_lock(1);
>
> in t1_lock_func. If you then make the session something roughly like
>
> s1: pg_advisory_xact_lock(1);
> s1: pg_advisory_xact_lock(2);
>
> s2: upsert t1 <blocking for 1>
> s1: pg_advisory_xact_unlock(1);
> s2: <continuing>
> s2: <blocking for 2>
> s1: insert into t1 values(1, 'someval');
> s1: pg_advisory_xact_unlock(2);
> s2: <continuing>
> s2: spec-conflict

Needed to be slightly more complicated than that, but not that much. See
the attached test. What do you think?

I think we should apply something like this (minus the WARNING, of
course). It's a bit complicated, but it seems worth covering this
special case.

Greetings,

Andres Freund