On 2020-Jan-31, Fujii Masao wrote:
> You're thinking to apply this change to the back branches? Sorry
> if my understanding is not right. But I don't think that back-patch
> is ok because it changes the documented existing behavior
> of pg_last_xact_replay_timestamp(). So it looks like the behavior
> change not a bug fix.
Yeah, I am thinking in backpatching it. The documented behavior is
already not what the code does. Do you have a situation where this
change would break something? If so, can you please explain what it is?
I think (and I said it upthread) a 100% complete fix involves tracking
two timestamps rather than one. I was thinking that that would be too
invasive because it changes XLogCtlData shmem struct ... but that struct
is private to xlog.c, so I think it's fine to change the struct. The
problem though is that the user-visible change that I want to achieve is
pg_last_xact_replay_timestamp(), and it would be obviously wrong to use
the new XLogCtlData field rather than the existing one, as that would be
a behavior change in the same sense that you're now complaining about.
So I would achieve nothing.
--
Álvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services