Consistent "region" subtag requirements for getTimeZones() and getWeekInfo()/getHourCycles() about proposal-intl-locale-info

[FrankYFTang]

I think there are three possible approaches we could consider for this issue

1. Keep the status quo
2. Change getWeekInfo() and getHourCycles() to also include the following step

If the unicode_language_id production of locale does not contain the ["-" unicode_region_subtag] sequence, return undefined.

3. Change getTimeZones() spec to return the value after adding likelySubtag

[sffc]

new Intl.Locale("en-US").getTimeZones()
// ["America/New_York", "America/Chicago", ...]

new Intl.Locale("en").getTimeZones()
// Current: undefined
// Anba suggestion: same as "en-US" based on likely subtags
// Shane suggestion: ["UTC"]
// Justin suggestion: []

[FrankYFTang]

We discussed this in 2023-09-07 TG2, with the following options:

1. maximize the locale to get the Region then to get the TimeZone
2. return ["UTC"]
3. return []
4. return undefined (AS IS in the current spec text)

attendees supprt keeping the current spec text that return undefined after consider the above options. Close this issue.

[sffc]

TG2 notes: https://github.com/tc39/ecma402/blob/master/meetings/notes-2023-09-07.md#consistent-region-subtag-requirements-for-gettimezones-and-getweekinfogethourcycles-71

[anba]

@sffc If it's not acceptable for getTimeZones() to use likely-subtags data to fill in missing regions, why is it acceptable that getWeekInfo() uses likely-subtags data to fill in missing regions?

[sffc]

Good question; the time zone impacts the actual numeric result of the data, whereas week info and hour cycle are about display preferences that have some impact on the layout but don't change the underlying quantity. It is therefore important to make a conscious choice about the time zone.

[anba]

And if we keep the current approach where the region subtag is only strictly required for getTimeZones(), but not for getWeekInfo(), should ICU4X follow suite and also fill in missing region subtags for unicode-org/icu4x#2999, or should this be done in the ICU4X caller?

The relevant ICU4C code is:

• Preprocessing to fill in missing region subtags: https://searchfox.org/mozilla-central/rev/6ee872a465eeb8769331c8cad74d136c9f13b67f/intl/icu/source/i18n/calendar.cpp#3862-3883
• Fallback to 001 if no region data was found: https://searchfox.org/mozilla-central/rev/6ee872a465eeb8769331c8cad74d136c9f13b67f/intl/icu/source/i18n/calendar.cpp#3920-3930

[sffc]

I anticipate that in places where a default can be reasonably derived, such as week info, likely subtags will be used during lookup.