Re: [TICTOC] [Sedate] [art] [Cbor] Two specifications on timestamps nearing WGLC

[Carsten Bormann <[e-mail]>]

When I need information about a paywalled document, I usually ask my students, who have a supernatural capability to divine information about such documents.

In this case, they were too lazy for actual divination, and I just got back a reference to a document apparently published by NIST: 

https://nvlpubs.nist.gov/nistpubs/Legacy/FIPS/fipspub4-1-1991.pdf

(This is missing the 1991 corrigendum, but it is a useful resource.
Delete the blank verso pages...)

Interestingly, I cannot find a prohibition of -00:00 (and I cannot find what would actually allow +00:00 either).  So the statement we got about -00:00 not being allowed by ISO 8601 apparently refers to a different revision…  (And I need to apologize to the RFC 3339 contributors who based their work on the 1988 revision, but then the RFC is also referencing ISO8601:2000, and that reference is also in draft-ietf-impp-datetime-04.)

My students also divined that ISO 8601:2000 appears to rule out using a minus sign for non-negative values.  One section that this can be inferred from is Section 5.1.1, but the result of that inference probably can be debated.  More importantly, Section 5.3.4.1 shows a fix that lets one infer +00:00 is actually allowed, while -00:00 is not.

Oh, and Library of Congress has this for us:

https://www.loc.gov/standards/datetime/iso-tc154-wg5_n0038_iso_wd_8601-1_2016-02-16.pdf

In that draft document (which is likely to reflect published versions up to its date, but may differ from ISO 8601:2019, the relevant sections that support the hypotheses about +00:00 being allowed and -00:00 not being allowed are 3.4.2 and 4.2.5.1.
Same section numbers for ГОСТ Р 7.0.64-2018, which seems to be ИСО 8601:2004 in Russian language.

Grüße, Carsten