Re: pg_locks display of speculative locks is bogus

Hi,

On 2020-02-11 12:24:50 -0800, Peter Geoghegan wrote:
> On Tue, Feb 11, 2020 at 12:03 PM Andres Freund <andres(at)anarazel(dot)de> wrote:
> > Doesn't seem great.
> >
> > It's trivial to put the xid in the correct place, but it's less obvious
> > what to do with the token? For master we should probably add a column,
> > but what about the back branches? Ignore it? Put it in classid or such?
>
> My vote goes to doing nothing about the token on the back branches.
> Just prevent bogus pg_locks output.
>
> Nobody cares about the specifics of the token value -- though perhaps
> you foresee a need to have it for testing purposes. I suppose that
> adding a column to pg_locks on the master branch is the easy way of
> resolving the situation, even if we don't really expect anyone to use
> it.

You can't really analyze what is waiting for what without seeing it -
the prime purpose of pg_locks. So I don't agree with the sentiment that
nobody cares about the token.

Greetings,

Andres Freund