DateTimeFormat `fractionalSecondDigits`: conflict between MDN and spec

[justingrant]

MDN shows that 0 is a valid value for the fractionalSecondDigits. The spec (and Chrome) throws when zero is passed for this option. Here's the docs:

I assume this was thought through and there's a good reason for zero being excluded in the 402 spec. Why was the reason?

FWIW, Temporal allows zero for this option, so regardless of whether zero is allowed or not, we should probably align across Temporal and 402. AFAIK, there's already a plan to expand this option to allow for nanoseconds precision, so as part of those changes I'd vote for allowing zero because it seems like a valid choice, but if you disagree then let's discuss. There's also an 'auto' option allowed on the Temporal side.

Here's the TS declaration of this on in Temporal:

fractionalSecondDigits?: 'auto' | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9;

Here's a repro to see the error

new Intl.DateTimeFormat(
  'en-us', 
  { hour: 'numeric', minute: 'numeric', second: 'numeric', fractionalSecondDigits: 0 }
).format(new Date())
// MDN: OK
// Spec and Chrome: RangeError: fractionalSecondDigits value is out of range.

[anba]

#347 initially always set fractionalSecondDigits, but I proposed to omit it when it's not present in the resolved date-time pattern. Then there was some confusion about how to handle undefined vs. 0, but that was resolved eventually. Later on Frank proposed to remove 0 from the set of allowed options altogether, presumably because that would have avoided this confusion about undefined vs. 0 in the first place.

---

If we were to allow 0, I strongly propose to implement this by setting 0 to undefined after reading "fractionalSecondDigits" from the options object. That means to specify this as:

Let value be ? GetNumberOption(options, "fractionalSecondDigits", 0, 3, undefined).
If value is 0, set value to undefined.

[ryzokuken]

This should be a simple normative PR, right? @justingrant would you like to take a stab at this?

[sffc]

Discussion from 2021-09-09: https://github.com/tc39/ecma402/blob/master/meetings/notes-2021-09-09.md#datetimeformat-fractionalseconddigits-conflict-between-mdn-and-spec

Conclusion: Seems reasonable to move forward with a spec change. Still some open questions from Anba and SFC.

[sffc]

My question on this issue is the timing of when we add 4-9. Should we add it into the Temporal proposal, or should we make the change sooner?