On Mon, 24 Apr 2023 23:47:05 GMT, Justin Lu <j...@openjdk.org> wrote:
>> This PR fixes an intermittent failure (that only occurs on Windows) in >> _DateFormatRegression.java_. >> >> With the integration of >> [JDK-8304982](https://bugs.openjdk.org/browse/JDK-8304982), >> _LocaleProviderAdapter.java_ now emits a compat warning when the class is >> loaded. This warning calls `ZonedDateTime.now()` during formatting. This >> call depends on the `TimeZone.ID` of the TimeZone returned from >> `TimeZone.getDefault()`. >> >> On Windows, the test class which DateFormatRegression extends will run the >> tests at random, (as opposed to running in the same order every time). When >> Test4089106() happens to be the first test ran, the static block of >> LocaleProviderAdapter will be executed with `TimeZone.setDefault()` set to a >> `SimpleTimeZone` with `id` as _FAKEZONE_. When LocaleProviderAdapter formats >> the compat warning ... and many calls later calls `ZoneRulesProvider >> getProvider(String zoneId)` with `zoneId` as _FAKEZONE_ the test fails with >> _java.time.zone.ZoneRulesException: Unknown time-zone ID: FAKEZONE_. >> >> In order to still test that `SimpleDateFormat.getTimeZone()` defaults to >> `TimeZone.getDefault()` we can create a `SimpleTimeZone` with a custom id >> rather than an invalid id. This way ZoneRulesProvider will not fail on the >> ID, but the `SimpleTimeZone` being tested is still not a "default" >> `TimeZone`. > > Justin Lu has updated the pull request incrementally with one additional > commit since the last revision: > > Minor cleanup / alternate custom tz Marked as reviewed by lancea (Reviewer). ------------- PR Review: https://git.openjdk.org/jdk/pull/13630#pullrequestreview-1400354287
