|
|
Am 09.08.2021 um 00:54 schrieb clipka:
> I hereby present the following alternative approaches as possible
> solutions:
My personal preferences lean heavily towards (A), with a perspective to
change over to (A.1) later if daylight savings time does indeed turn out
to be a problem.
(A) explicitly includes provisions to let people know that the `local`
mode may still be subject to change, for this exact reason.
The reason for any DST issues would be that, in order for stuff to work
nicely, the time zone of the timestamp "zero point" (2000-1-1 00:00)
probably needs to match the time zone applicable for (a) the point in
time represented by the timestamp, or (b) the point in time at which the
scene is parsed (and I'm not sure which of the two). Either way, since
2000-1-1 is outside the DST period for most time zones, there is reason
to expect things to break down when invoking `datetime(now)` during the
DST period, since the DST time zone typically doesn't match the regular
time zone.
Since UTC does not have a DST variant, this is a non-issue if our
numeric timestamps are always UTC-based.
Post a reply to this message
|
|