On a somewhat disappointing correspondence (was: max_standby_delay considered harmful)

Simon Riggs <simon(at)2ndQuadrant(dot)com> wrote:

I've refrained from comment on max_standby_delay because I have
neither read the patch nor am likely to be an early adopter of HS;
however, as a potential eventual user I have to say that the
semantics for this GUC proposed by Simon seem sane and useful to me.

Certainly the documentation would need to be clear on the pitfalls
of using something other than 0 or -1, and there were technical
issues raised on the thread outside the scope of the semantics of
the GUC, but the issues around clock sync and transfer time ring of
FUD. We sync our central router to a bank of atomic clocks around
the world, and sync every server to the router -- if a server drifts
we would have much bigger problems than this GUC would pose, so we
monitor that and make loud noises should something drift.

Are there other controls that would be useful? Undoubtedly. Should
they be added to 9.0? I'm not in a position to say. I don't see
the point of ripping out one potentially useful control, which
*might* be sufficient for 9.0 because someone might choose to use it
inappropriately. Just make sure it's documented well enough.

> Yesterday you berated me for unstable software. Today you oppose
> my promise to fix that. Why is it, we all wonder, is it that you
> oppose everything I say and do?

Robert strikes me as a top-notch project manager, and his comments
struck me as totally in line with someone wearing that hat.

-Kevin