Re: [art] [saag] Date formats: RFC3339 vs. RFC 5322

[Carsten Bormann <[e-mail]>]

On 2021-04-14, at 22:53, Mark Baushke (ietf) <[e-mail]> wrote:
> 
> Twelve digits after the decimal point expresses pico seconds which is just silly and unrealistic at present, but micro seconds seem plausible in many contexts. If new standards are forward looking, then the expression should possibly be considered.
> (I do not see a need to consider femto, atto, zepto, or yacto seconds at the present time.)

I don’t know why this discussion is continuing on the SAAG list, but let me add one factoid anyway:

The CBOR tag defined in draft-bormann-cbor-time-tag supports picosecond resolution [1] because the Haskell date/time libraries need that (which in turn probably just try to be one step ahead of UNIX nanosecond resolution).  (The CBOR tag also has femtosecond and attosecond resolution, because these are still cheap with 64-bit integers.  Zeptoseconds, yoctoseconds, and anything else down to and beyond Planck units can be added later based on bignums.)

A light-nanosecond is ~ 300 mm (~ a foot), a light-picosecond still is ~ 0.3 mm (~   12 thou), which is well beyond the tolerances to which e.g. connectors can be manufactured, so while these are not easy to achieve as date-time precisions, as durations (time differences) they are in wide use (as are femtoseconds and even attoseconds).

RFC 3339 has no problem with decimal fractions at all, so there is no need for invention as long as we talk about text strings for date-times.

Grüße, Carsten

[1]: https://www.ietf.org/archive/id/draft-bormann-cbor-time-tag-04.html#name-keys-3-6-9-12-15-18