Re: [Cbor] draft-ietf-cbor-date-tag-02 - handling of time zone offsets

[Juergen Schoenwaelder <[e-mail]>]

On Wed, Jul 01, 2020 at 06:23:30PM +0200, Carsten Bormann wrote:
> On 2020-07-01, at 17:51, Michael Richardson <[e-mail]> wrote:
> > 
> > The failure of the Nipponese to declare war on the USA on December 8, 1941
> > (they date of the attack on Pearl Habou^Hr in Japan...) is perhaps a better
> > example of failures to get the time zone right.
> 
> … and then there is V-E Day which is on 1945-05-08 everywhere except in some countries usually more or less loosely related to the Soviet Union, where it is on 1945-05-09 (even though the actual capitulation document was signed on 21:20 Central European Summer Time, which I think was 22:20 in Moscow which didn’t follow DST at the time…  
> No time left to set up parades, so -05-09 it was :-).
> 
> I’m citing this as an example where a calendar date was chosen entirely by fiat, not based on the exact time of the actual event as adjusted for a specific place/time zone.  It does make sense to separate calendar dates from day-length periods anchored on a time scale, and this document is about representing calendar dates in CBOR.
>

You often associate an event with a date. If you ignore timezone, then

 (1) two systems may associate the same event with different dates and
 (2) comparisons of date values may lead to "wrong" results.

Since number of days since 1970-01-01 seems to be calculated in UTC
(the document refers to the POSIX epoch which seems to be in UTC)

 (3) a 'date' value and a 'number of days since 1970-01-01' value can
     be "inconsistent".

I think this deserves a clear warning. The -03 version has text in the
security considerations section scoped to access control decisions but
the fact that comparisons of date values and number of days since
1970-01-01 values may be surprising is not limited to access control,
i.e., I prefer to see this explicitly discussed in section 2.1,
perhaps including advice what should be done if proper comparison of
date values is desirable.

/js

-- 
Juergen Schoenwaelder           Jacobs University Bremen gGmbH
Phone: [phone]         Campus Ring 1 | 28759 Bremen | Germany
Fax:   [phone]         <https://www.jacobs-university.de/>