1950162 - Temporal API, wrong results for yearOfWeek/weekOfYear w/ calendar:gregory

Status: RESOLVED FIXED

(Core :: JavaScript: Standard Library, defect, P3)

Description

[Adam Shaw]

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/132.0.0.0 Safari/537.36

Steps to reproduce:

Test on Firefox nightly 137.0a1 (2025-02-24) (aarch64)

const lastDay2029 = Temporal.PlainDate.from('2029-12-31[u-ca=gregory]');

/*
EXPECT: 1
ACTUAL: 53
*/
console.log(lastDay2029.weekOfYear)

/*
EXPECT: 2030
ACTUAL: 2029
*/
console.log(lastDay2029.yearOfWeek)

Actual results:

See above

Expected results:

See above

Comment 1

[Adam Shaw]

Originally reported here: https://github.com/fullcalendar/temporal-polyfill/issues/53

Comment 2

[BugBot [:suhaib / :marco/ :calixte]]

The Bugbug bot thinks this bug should belong to the 'Core::JavaScript: Standard Library' component, and is moving the bug to that component. Please correct in case you think the bot is wrong.

Comment 3

[André Bargull [:anba]]

Can you file an issue at https://github.com/tc39/proposal-temporal/ to clarify the spec? If it is expected that the Gregorian calendar uses exactly the same week numbering system as the ISO-8601 calendar, the spec should require this behaviour. Thanks!

Comment 4

[Adam Shaw]

This is getting discussed here: https://github.com/tc39/proposal-temporal/issues/3096

Comment 5

[Adam Shaw]

Hi André, hope you are well. After digging further, I've realized the problem isn't that gregorian week numbers should/shouldn't be following iso8601, but rather the gregorian week numbers that Firefox is generating are buggy on their own right. Please see this recreation code:

d = Temporal.PlainDate.from('<insert PlainDate string>');
console.log(d.yearOfWeek + ', ' + d.weekOfYear);

Results:

PlainDate string	Day of Week	Ref Impl	Firefox
2029-12-30[u-ca=gregory]	7 - Sun	2029, 52	2029, 52
2029-12-31[u-ca=gregory]	1 - Mon	2030, 1	2029, 53
2030-01-01[u-ca=gregory]	2 - Tue	2030, 1	2030, 1
2030-01-02[u-ca=gregory]	3 - Wed	2030, 1	2030, 1
2030-01-03[u-ca=gregory]	4 - Thu	2030, 1	2030, 1
2030-01-04[u-ca=gregory]	5 - Fri	2030, 1	2030, 1
2030-01-05[u-ca=gregory]	6 - Sat	2030, 1	2030, 1
2030-01-06[u-ca=gregory]	7 - Sun	2030, 1	2030, 1
2030-01-07[u-ca=gregory]	1 - Mon	2030, 2	2030, 2

Note that Firefox is amazingly determining that week 2030, 1 only has 6 days in it!

I tried tracing the code from here -> here, and I see this is offloaded to ICU4X. All the configuration looks correct to me (firstWeekday = Monday, minWeekDays = 1), so could it be that this is a bug in ICU4X?

Comment 6

[Adam Shaw]

Err, maybe ICU4X isn't necessarily "buggy" but it's opinionated about week numbers spanning as far as possible until year-end.

Comment 7

[Adam Shaw]

This was discussed in the 3/6/2025 Temporal champions meeting. It's likely a bug in ICU4X not conforming to the latest CLDR. I've opened a ticket here: https://github.com/unicode-org/icu4x/issues/6247

Comment 8

[André Bargull [:anba]]

Attachment: Bug 1950162: Return undefined for weekOfYear and yearOfWeek for Gregorian. r=#spidermonkey-reviewers!

Return undefined per the discussions in https://github.com/tc39/proposal-temporal/issues/3096
and https://github.com/tc39/proposal-intl-era-monthcode/issues/15.

Comment 9

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/efd59ac77d7b
https://hg.mozilla.org/integration/autoland/rev/572d1620d5bd
Return undefined for weekOfYear and yearOfWeek for Gregorian. r=spidermonkey-reviewers,dminor

Comment 10

[Sandor Molnar[:smolnar]]

https://hg.mozilla.org/mozilla-central/rev/572d1620d5bd