* We handle timezones by converting to UTC time during parse
* Ends up leaving wall-times that look like locatimes,
but are UTC
* Causes times to advance by zone-offset every read/write
to a timezone aware database (postgresql-socket3)
* To prevent this we will track whether a date will be in UTC
or is a local time. When writing out UTC timestamps we add a Z to
the end
* During comparison, we always convert to UTC to compare,
we assume zoneless timestamps are localtimes
There was a lot of personal debate about how to do this. I think a
better answer would be to fully replace clsql-date/times with the
local-time library which has a more comprehensive handling of dates
and times.
re ADWolf:#1408