1496584 - Implement Intl.DateTimeFormat.prototype.formatRange and .formatRangeToParts

Status: RESOLVED FIXED

(Core :: JavaScript: Internationalization API, enhancement, P3)

Description

[Jason Orendorff [:jorendorff]]

ECMA 402 proposal, currently stage 2:
  https://github.com/fabalbon/proposal-intl-DateTimeFormat-formatRange

Comment 1

[Daniel Ehrenberg]

This is now Stage 3, and Chrome has implemented it, planning on shipping soon.

Comment 2

[André Bargull [:anba]]

I have a local patch for this proposal, but the following ICU bugs/issues prevent implementing/shipping it anytime soon.

• https://unicode-org.atlassian.net/browse/ICU-20705
• https://unicode-org.atlassian.net/browse/ICU-20706
• https://unicode-org.atlassian.net/browse/ICU-20707
• https://unicode-org.atlassian.net/browse/ICU-20710

Comment 3

[André Bargull [:anba]]

Attachment: Bug 1496584: Add support for Intl.DateTimeFormat.prototype.formatRange proposal. r=jwalden!

Comment 4

[André Bargull [:anba]]

Uploaded a draft patch, which due to the reasons outlined in comment #2, can't land.

Comment 5

[Shane Carr]

Fixes for the linked issues are landing in ICU 67, which is being released today.

Is there anything else blocking the formatRange proposal in SpiderMonkey?

Comment 6

[André Bargull [:anba]]

I've performed some cursory testing and didn't find any obvious blockers. I did find two ICU bugs [1,2], but those are probably not hard blockers.

[1] Mismatched hour skeleton and hc Unicode extension keys result in different outputs depending on whether or not a range is present.
[2] ICU doesn't seem to fully implement step 2 from https://unicode.org/reports/tr35/tr35-dates.html#intervalFormats, for example "de.xml" contains

<intervalFormatItem id="H">
  <greatestDifference id="H">HH–HH 'Uhr'</greatestDifference>
</intervalFormatItem>

but ICU doesn't seem to use this interval format when the hour skeleton character is "k" instead of "H".

Comment 7

[André Bargull [:anba]]

Attachment: Bug 1496584 - Part 1: Move FormattedValueToString into CommonFunctions.h. r=jwalden!

That way this function can be reused for the date range formatting.

Comment 8

[André Bargull [:anba]]

Attachment: Bug 1496584 - Part 2: Include previously disabled date-interval formatters. r=jwalden!

Depends on D79317

Comment 9

[André Bargull [:anba]]

Attachment: Bug 1496584 - Part 3: Split DateTimeFormatLocale from NewUDateFormat. r=jwalden!

Move the locale computation into a separate function, so we can reuse it in
a later patch.

Depends on D79318

Comment 10

[André Bargull [:anba]]

Attachment: Bug 1496584 - Part 4: Rename the estimated-size member of DateTimeFormatObject. r=jwalden!

Later patches will add the interval formatter, whose size also needs to be
tracked.

Depends on D79319

Comment 11

[André Bargull [:anba]]

Attachment: Bug 1496584 - Part 5: Split skeleton computation from toBestICUPattern. r=jwalden!

The date interval formatter requires a skeleton for its input, so we need to
store the skeleton string in addition to the pattern string. Prepare for this
change by splitting the skeleton computation from toBestICUPattern.

Depends on D79320

Comment 12

[André Bargull [:anba]]

Attachment: Bug 1496584 - Part 6: Add support for Intl.DateTimeFormat.prototype.formatRange proposal. r=jwalden!

Comment 13

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/ce4987951537
Part 1: Move FormattedValueToString into CommonFunctions.h. r=jwalden
https://hg.mozilla.org/integration/autoland/rev/509364db7c0d
Part 2: Include previously disabled date-interval formatters. r=jwalden
https://hg.mozilla.org/integration/autoland/rev/02cdd5e863ae
Part 3: Split DateTimeFormatLocale from NewUDateFormat. r=jwalden
https://hg.mozilla.org/integration/autoland/rev/f354edbe8be6
Part 4: Rename the estimated-size member of DateTimeFormatObject. r=jwalden
https://hg.mozilla.org/integration/autoland/rev/25d650de6d6c
Part 5: Split skeleton computation from toBestICUPattern. r=jwalden
https://hg.mozilla.org/integration/autoland/rev/373d4c8b47cd
Part 6: Add support for Intl.DateTimeFormat.prototype.formatRange proposal. r=jwalden

Comment 14

[Razvan Maries]

https://hg.mozilla.org/mozilla-central/rev/ce4987951537
https://hg.mozilla.org/mozilla-central/rev/509364db7c0d
https://hg.mozilla.org/mozilla-central/rev/02cdd5e863ae
https://hg.mozilla.org/mozilla-central/rev/f354edbe8be6
https://hg.mozilla.org/mozilla-central/rev/25d650de6d6c
https://hg.mozilla.org/mozilla-central/rev/373d4c8b47cd

Comment 15

[Jeff Walden [:Waldo]]

Given the workarounds for the ICU issues in the patch, is there a need for this to be nightly-only any more? The patches landed with that configuration, but it's not clear to me what prevents a followup bug making this not merely nightly-only at this point.

Comment 16

[André Bargull [:anba]]

The spec has two issues I'd like to see resolved before letting this feature ride to release:

1. There's https://github.com/tc39/proposal-intl-DateTimeFormat-formatRange/issues/21 which means the spec as-is doesn't do the correct thing at all. I've added some comments how it could be fixed, but nobody seems to have looked into that one yet.
2. And I'm not sure how people feel about https://github.com/tc39/proposal-intl-DateTimeFormat-formatRange/issues/22. Proposal champions and reviewers need to chime in there, too.