Andrew Dunstan <andrew@dunslane.net> writes:
> Tom's idea of a log_timezone seems to make sense.
Here's a preliminary patch for this --- no docs yet, but code is all
there. Seems to work OK. The patch is larger than it'd really have to
be because I chose to rename global_timezone to session_timezone;
I thought it'd be confusing to have something called "global_timezone"
that actually has a narrower scope than "log_timezone".
This wouldn't do for backpatching of course. My inclination for a
backpatch is to just take %Z out of the strftime formats for Windows,
as we did in several places already.
Comments?
regards, tom lane