Normative: Change the hourCycle default logic

[FrankYFTang]

While hour12 is false, let hourCycle to be alwasyt 'h23' but not 'h24'.

The current logic is not reasonable. We see no region in the CLDR use h24 hour cycle as default. While we set a hour12: false on a 12 hour system region (which use 1:00 - 12:59), we should set to h23 instead of h24

We need to make this change
so while hour12 is false on 'en' locale, it will be set to h23 instead of h24
but also while hour12 is true on 'ja' locale, it will be set to h11 but not h12.

Address #402

[FrankYFTang]

notice grep "<hours.*allowed" ./cldr/common/supplemental/supplementalData.xml show possibility of h H or K but no possibility of k in the latest CLDR and it is unlikely that will happen in the future from my understanding.

[FrankYFTang]

PTAL I took anba's suggestion and make it data driven.

[sffc]

TG2 agreement: https://github.com/tc39/ecma402/blob/master/meetings/notes-2023-04-06.md#normative-change-the-hourcycle-default-logic-758

[FrankYFTang]

somehow no one ever present this PR to TG39 in May 2023 or July 2023 meeting. I add it to https://github.com/tc39/agendas/blob/main/2023/09.md

[FrankYFTang]

we need test262 tests for this PR

[FrankYFTang]

@anba - any suggetion how should we write test262 test to test this PR?

[anba]

@anba - any suggetion how should we write test262 test to test this PR?

https://github.com/tc39/test262/blob/main/test/intl402/DateTimeFormat/prototype/resolvedOptions/hourCycle-default.js needs to be updated to reflect the PR changes. The test is already marked as locale-specific, so we could simply change it to:

let locales = ["en", "fr", "it", "ja", "zh", "ko", "ar", "hi"];

for (let locale of locales) {
  let hcDefault = new Intl.DateTimeFormat(locale, {hour: "numeric"}).resolvedOptions().hourCycle;
  assert(
    hcDefault === "h11" || hcDefault === "h12" || hcDefault === "h23" || hcDefault === "h24",
    "hcDefault is one of [h11, h12, h23, h24]"
  );

  let hour12 = new Intl.DateTimeFormat(locale, {hour: "numeric", hour12: true}).resolvedOptions().hourCycle;
  assert(hour12 === "h11" || hour12 === "h12", "hour12 is one of [h11, h12]");

  let hour24 = new Intl.DateTimeFormat(locale, {hour: "numeric", hour12: false}).resolvedOptions().hourCycle;
  assert(hour24 === "h23" || hour24 === "h24", "hour24 is one of [h23, h24]");

  if (hcDefault === "h11" || hcDefault === "h12") {
    assert.sameValue(hour12, hcDefault, "hour12 matches hcDefault");
  } else {
    assert.sameValue(hour24, hcDefault, "hour24 matches hcDefault");
  }

  // 24-hour clock uses the "h23" format in all locales.
  assert.sameValue(hour24, "h23");

  // 12-hour clock uses the "h12" format in all locales except "ja".
  assert.sameValue(hour12, locale === "ja" ? "h11" : "h12");
}

[ryzokuken]

@FrankYFTang @anba is merging this PR blocked on the tests?

[ryzokuken]

2023-09-26: This PR achieved TG1 consensus

[FrankYFTang]

This PR reach consensus on 2023-09-26 meeting

[FrankYFTang]

I believe the PR need to merge first before merging the test, right?

[FrankYFTang]

@FrankYFTang @anba is merging this PR blocked on the tests?

The test is blocked by the merge of this PR tc39/test262#3885

[anba]

Two small issues in the description text, otherwise LGTM!

[jufemaiz]

🎉

[jufemaiz]

Can I assume that this will form part of a changeset for the 10th edition?

Ref: https://ecma-international.org/publications-and-standards/standards/ecma-402/