leonbloy@sinectis.com.ar writes:
> Tom Lane wrote:
>> I'll bet there is some bit of internal state somewhere that affects
>> the results. It could be inside libc, or it could be in Postgres.
> postgres, I would tend to think...
> For one thing I've just found out: the 'histeresis' effect occurs
> only WITHIN A CONNECTION:
Doesn't prove a thing. The libc state I'm thinking of would be
recomputed on first use within a given process, ie, per backend.
So we still have no idea if the problem is libc's or Postgres'.
Given that the problem seems to be specific to your timezone,
and Postgres itself knows very little about timezones, a libc bug
is well within the realm of possibility.
Is anyone else able to reproduce this misbehavior at a DST boundary?
regards, tom lane
