[Numpy-discussion] timezones and datetime64

[Daniele Nicolodi]

On 04/04/2013 01:27, Pauli Virtanen wrote:
>
> Probably also TAI and UTC/Posix.
> 
> Converting from one format to the other is problematic since
> all of them (except TAI afaik) require looking things up in
> regularly updated databases. Not only restricted to conversions,
> but also arithmetic, `b - a`. Affects also UTC/Posix via leap
> seconds --- this probably doesn't usually matter, but if we want
> to be strict, it's not good to ignore the issue.

I was about to point out the same issue with UTC.

I'm not aware of any library that handles the conversion from UTC to
TAI. I would like to know if there is one. Furthermore, UTC cannot be
computed unambiguously for times in the future because the leap second
insertion is not scheduled regularly but it is based on the comparison
between UT1 and UTC times.

I think that generally the issue is not relevant for any practical use
of a timebase: there are not many applications requiring sub-second
accuracy over many years periods. Solving it correctly (having an
internal representation in TAI time and converting to UTC and then to
local time for user representation) is complex and potentially confusing
for the users (i don't know how many scientists know what TAI is). I
think it may be neglected in the code but stated in the documentation.

Cheers,
Daniele