The following layout test is failing on Windows js/date-constructor.html Failure output: --- /cygdrive/c/Projects/WebKit/OpenSource/WebKitBuild/Release/bin32/layout-test-results/js/date-constructor-expected.txt +++ /cygdrive/c/Projects/WebKit/OpenSource/WebKitBuild/Release/bin32/layout-test-results/js/date-constructor-actual.txt @@ -38,7 +38,9 @@ PASS new Date(Date.parse('1 Mar 1000 GMT')).toGMTString() is "Sat, 01 Mar 1000 00:00:00 GMT" PASS testStr is "1234567" PASS testStr is "1234567" -PASS leapTestResult is true +WARN: shouldBe() expects string arguments +FAIL new Date(August 15, 101).getDate() should be 15. Threw exception SyntaxError: Unexpected number '15'. Expected ')' to end a argument list. +FAIL leapTestResult should be true. Was false. PASS successfullyParsed is true TEST COMPLETE
<rdar://problem/19617672>
Could be due to Bug 140917.
More pointedly: - `new Date("January 15, 2001")` prints "Mon Jan 15 2001 00:00:00 GMT-0800 (Pacific Standard Time)" (expected) - `new Date("August 15, 2001")` prints "Tue Aug 14 2001 00:00:00 GMT-3100 (Pacific Daylight Time)" (unexpected) It seems that we're not checking for a negative diff in calculateDSTOffset on Windows.
Created attachment 338176 [details] Patch
Created attachment 338178 [details] Patch
Comment on attachment 338178 [details] Patch R=me.
Comment on attachment 338178 [details] Patch Clearing flags on attachment: 338178 Committed r230774: <https://trac.webkit.org/changeset/230774>
All reviewed patches have been landed. Closing bug.