> With the U.S. Government poised to tinker with Daylight
> Saving Time, I have a couple questions.
>
> 0) Does PG rely on system-level TZ info or will specific
> updates to PG be required? I'm interested in *nix but the
> Windows folks are probably interested, too.
Pre-8.0 relies on the system-level TZ info. So you'd need an OS update.
8.0 and later comes with it's own timezone database (which is synced
from upstream, it's not locally created). So you'd need a PostgreSQL
update for this (you could just update the timezone files and not the
rest if you wanted to, but I would assume a new minor postgresql version
wluld be put out)
> 1) However a change is implemented, what will be the effect
> on queries for historical data? Will the localtime be correct
> for all dates whether before or after the change?
The timezone database as imported into 8.0, and as present on many
(most?) Unix systems already handles the fact that DST rules have
changed over time. It will apply the timezone rule that was valid at the
time in question.
> Perhaps we will still dodge this bullet. But with the House
> and Senate recently reaching an agreement, the change is
> appearing ever more likely.
This happens all the time around the world. It's just been a while in
the US :-)
(For example, not too many years ago a lot of countries in europe
changed their rules to be same across the EU)
//Magnus