Problem saving datetime to file and reading it back for a calculation

Chris Angelico rosuav at gmail.com
Sat Oct 10 22:02:15 EDT 2020


On Sun, Oct 11, 2020 at 12:57 PM Cameron Simpson <cs at cskk.id.au> wrote:
> Personally I strongly dislike using datetimes for computation or as the
> basis for time record keeping, essentially because of the timezone issue
> but also because the human calendar is a complex disaster of illfitting
> units (days in a year? variable; days in a month? variable; that table
> of days per unit? variable depending on your time in history).
>
> Instead, I always try to work in POSIX timestamps, an absolute number of
> seconds since midnight, 1 January 1970 GMT. You can always do arithmetic
> directly between these in seconds, then convert for presentation
> purposes whenever.
>

Absolutely agree, with the annoying exception of recurring events. If
a human says "this happens every Monday at 2pm", then the human
expects it to recur every Monday at 2pm, not every 604800 seconds. The
difference shows up when the conversion between UTC and local time
changes - most commonly when Daylight Saving Time starts or ends...

(And I guess if you care about leap seconds, then Unix time would be
inappropriate there too. But I rather doubt that most of us are
bothered by that.)

ChrisA


More information about the Python-list mailing list