On Jul16, 2011, at 22:55 , Tom Lane wrote:
> Florian Pflug <fgp@phlo.org> writes:
>> Btw, if we touch that, I think we should think about providing some way
>> to detect when a backend fails to apply a value.
>
> Hm, maybe, but keep in mind that there are valid reasons for a backend
> to ignore a postgresql.conf setting --- in particular, it might have a
> local override from some flavor of SET command. So I don't think we'd
> want the flag to have the semantics of "this backend is actually *using*
> the value";
Yeah, the flag would simply indicate whether a particular backend
encountered an error during config file reload or not.
Actually being able to inspect other backend's GUCs would be nice, but
is way beyond the scope of this of course.
> and yet, if that's not what it means, people could still be
> confused.
Hm, if it's called "cfgfile_valid" or a prettier version thereof I
think the risk is small.
> There might be some implementation gotchas as well. I'm not
> sure offhand how thoroughly the GUC code checks a value that is being
> overridden.
If it doesn't, then what happens when the overriding scope ends, and
the value reverts (or attempts to revert) to its default?
best regards,
Florian Pflug