incorporate RFC 9557 time zone extension into 'date-time' format

[gregsdennis]

What kind of change does this PR introduce?

Enhancement

Issue & Discussion References

• Closes ✨ Proposal: date-time format to accept those in RFC 9557 #1572

Summary

Extends the date-time format by incorporating time zone information from RFC 9557.

Does this PR introduce a breaking change?

No Yes? date-time strings that would have previously failed would now pass.

[karenetheridge]

Isn't this a breaking change, if strings that were invalid before are now considered valid?

[gregsdennis]

That's what I was thinking, too. Someone mentioned it wasn't, and I failed to research it.

[Relequestual]

I agree this would be a breaking change.

Additionally, note we specify the duration format from ISO 8601 as per RFC 3339 Appendix A.

RFC 3339 Appendix A notes...

This information is based on the 1988 version of ISO 8601. There may
be some changes in the 2000 revision.

I'm positive we discussed before and opted to not reference ISO directly because of the paywall nature of the standard.

[gibson042]

Comments from someone familiar with RFC 3339, RFC 9557, and ISO 8601.

[gibson042]

This text is not clear to me... is the intent to allow a subset of the RFC 9557 grammar that includes time-zone but not suffix-tag? If so, I think that should be more explicit:

Suggested change

	valid representation according to the "date-time" ABNF rule in RFC 3339,
	optionally with the "time-zone" rule in RFC 9557
	valid representation of the "date-time" ABNF rule in RFC 3339 or such a
	string that is immediately followed by a valid representation of the
	"time-zone" rule in RFC 9557

[gibson042]

ISO 8601 does not use ABNF, or have "rules" as such.

Suggested change

	representation according to the "duration" ABNF rule in ISO 8601
	representation according to the "duration" ABNF rule in RFC 3339 Appendix A

[gibson042]

Suggested change

	format is from the ISO 8601 ABNF as given in Appendix A of RFC 3339.
	format is from ISO 8601 as formalized into ABNF by RFC 3339 Appendix A.

[awwright]

Yeah, I read that the update to date-time has not changed the syntax, the only update around date-time is to align Z with the common usage of "no time zone".

[jdesrosiers]

Isn't this a breaking change, if strings that were invalid before are now considered valid?

I don't consider this a breaking change because format allows partial validation. If strict validation was required, it would be a different story. Every value that passed in 2020-12 would still pass with these changes. I think this change just makes what was previously considered a strict validation implementation into a partial validation implementation.