Bug 4975 – For default Russian date format, results of toLocalizedString cannot be parsed, resulting in a regression test failure

• Component: JavaScriptCore
• Status: NEW

[Alexey Proskuryakov]

If the system primary language is Russian, Date.parse() fails on results of toLocalizedString(), making 
ecma_3/Date/15.9.5.6.js be reported as a regression.

Failure messages were:
--> (Date.parse(now.toLocaleDateString()) - (midnight(now)).valueOf()) == 0 = false FAILED! expected: 
true
--> Date.parse(Thu Jan 01 1970 03:00:00 GMT+0300).toLocaleDateString()) = NaN FAILED! expected: 
-10800000
--> Date.parse(Thu Jan 01 1970 06:00:00 GMT+0300).toLocaleDateString()) = NaN FAILED! expected: 
-10800000
--> Date.parse(Mon Jan 01 1900 03:00:00 GMT+0300).toLocaleDateString()) = NaN FAILED! expected: 
-2208999600000
--> Date.parse(Mon Jan 01 1900 00:00:00 GMT+0300).toLocaleDateString()) = NaN FAILED! expected: 
-2208999600000
--> Date.parse(Sat Jan 01 2000 03:00:00 GMT+0300).toLocaleDateString()) = NaN FAILED! expected: 
946674000000
--> Date.parse(Sat Jan 01 2000 00:00:00 GMT+0300).toLocaleDateString()) = NaN FAILED! expected: 
946674000000
--> Date.parse(Tue Feb 29 2000 03:00:00 GMT+0300).toLocaleDateString()) = NaN FAILED! expected: 
951771600000
--> Date.parse(Tue Feb 29 2000 02:59:59 GMT+0300).toLocaleDateString()) = NaN FAILED! expected: 
951771600000
--> Date.parse(Tue Feb 29 2000 00:00:00 GMT+0300).toLocaleDateString()) = NaN FAILED! expected: 
951771600000
--> Date.parse(Sat Jan 01 2005 03:00:00 GMT+0300).toLocaleDateString()) = NaN FAILED! expected: 
1104526800000
--> Date.parse(Sat Jan 01 2005 02:59:59 GMT+0300).toLocaleDateString()) = NaN FAILED! expected: 
1104526800000
--> Date.parse(Sat Jan 01 2005 00:00:00 GMT+0300).toLocaleDateString()) = NaN FAILED! expected: 
1104526800000

[Alexey Proskuryakov]

Created attachment 3889 [details]
test case

Steps to reproduce with this test case:

1. Set Region in International control panel (Formats tab) to Russia
2. Open the document in Safari (I was running current ToT; same with 412.2)

Expected results: two alerts, first with a Russian-lnaguage date, second with a
number
Actual results: two alerts, first with a Russian-lnaguage date, second with NaN

[Alexey Proskuryakov]

(In reply to comment #0)
> If the system primary language is Russian

Correction: this probably has nothing to do with primary language, just the regional format.

[Alexey Proskuryakov]

The current source from Mozilla now says:

----------------------------------------------------------------
    Note: This test isn't supposed to work with a non-English locale per spec.
----------------------------------------------------------------

With WebKit, it also won't work if the user changes the date format International Preferences. Maybe this 
test just needs to be removed or disabled?

[Alexey Proskuryakov]

Here's where Mozilla people have decided to keep this test: <https://bugzilla.mozilla.org/show_bug.cgi?
id=282339>.

Anyway, having a test for something that is guaranteed to fail for most people in the world is strange.

[Joost de Valk (AlthA)]

Adding Regression keyword.

[Joost de Valk (AlthA)]

Upping to P1 because this is a regression.

[Alexey Proskuryakov]

In fact, I have no reason to believe it is a regression, although it is reported as such by the tests. Moving the priority/severity down; removing the keyword.

[Darin Adler]

I think we should probably disable this test unless we have some way to set the locale in the test.