| b49c7c3f6a74d80580682b53457438a9e95f961d |
|
10-Feb-2017 |
Joachim Sauer <jsauer@google.com> |
Add tests for GregorianCalendar.
Add tests to GregorianCalendar.from(ZonedDateTime) and
GregorianCalendar.toZonedDateTime().
Bug: 28832222
Test: vogar libcore.java.util.GregorianCalendarTest
Change-Id: I823e75e17e0110cb5d92c1de40e0470899c99765
|
| ebffddd8bf06a0d9c862e222972544b11788aa00 |
|
09-Feb-2017 |
Tobias Thierer <tobiast@google.com> |
Tests for Calendar APIs from OpenJDK8u60
This covers:
- GregorianCalendar.{from,to}ZonedDateTime()
- Calendar.toInstant()
- Calendar.getAvailableCalendarTypes()
Notes:
- The supported calendar types may change in a follow-up CL
(bug 35138741). Currently, only "gregory" is listed even
though Calendar.Builder.build() still attempts to use
"buddhist" (this inconsistency is a bug).
- Calendar.Builder is not covered by these tests
Test: CalendarTest, GregorianCalendarTest
Bug: 34923731
Bug: 29935305
Change-Id: Icfa87f465941e63f11f4ef6c35f82cdbcfd313fb
|
| a4750c7d39fb853458c8deb1d0d7da780b29b15a |
|
15-Apr-2016 |
Neil Fuller <nfuller@google.com> |
Explicitly set locale to en_US for tests that require it
Previously these tests would use the device setting. The follow tests
broke if the device was set to en_UK:
libcore.java.text.SimpleDateFormatTest#testTimeZoneNotChangedByParse
org.apache.harmony.tests.java.util.GregorianCalendarTest#test_getLeastMaximumI
libcore.java.util.GregorianCalendarTest#test_getWeeksInWeekYear
The GregorianCalendar tests broke because the locale influences the
"start of week".
The SimpleDateFormat test broke because HST is not an ICU translation for
en_UK but is for en_US.
Bug: 28212178
(cherry picked from commit 3819290afba2199d4ddf5da88e5b3717b59863f6)
Change-Id: I15f6e0f0133a876b0d4173dacef8fad7506f52ab
|
| 3819290afba2199d4ddf5da88e5b3717b59863f6 |
|
15-Apr-2016 |
Neil Fuller <nfuller@google.com> |
Explicitly set locale to en_US for tests that require it
Previously these tests would use the device setting. The follow tests
broke if the device was set to en_UK:
libcore.java.text.SimpleDateFormatTest#testTimeZoneNotChangedByParse
org.apache.harmony.tests.java.util.GregorianCalendarTest#test_getLeastMaximumI
libcore.java.util.GregorianCalendarTest#test_getWeeksInWeekYear
The GregorianCalendar tests broke because the locale influences the
"start of week".
The SimpleDateFormat test broke because HST is not an ICU translation for
en_UK but is for en_US.
Bug: 28212178
Change-Id: Id4c9ca48dcc821e8f7c6af22d7e8a9fbce57d62c
|
| 869ce9a33446e07f665bd5cdf4b75a73ad314d39 |
|
26-Mar-2016 |
Shubham Ajmera <shubhamajmera@google.com> |
Add unit tests for java.util.Calendar
and GregorianCalendar
Tests added for:
Calendar.getWeekYear()
Calendar.getWeeksInWeekYear()
Calendar.isWeekDateSupported()
Calendar.setWeekDate(int, int, int)
GregorianCalendar.getWeekYear()
GregorianCalendar.getWeeksInWeekYear()
GregorianCalendar.isWeekDateSupported()
GregorianCalendar.setWeekDate(int, int, int)
Bug: 27330986
(cherry-picked from commit 10a1d6caad617432420db593960829e47b38f761)
Change-Id: I24faf062ec966c097ba84a09292654d1e178e9a4
|
| 10a1d6caad617432420db593960829e47b38f761 |
|
26-Mar-2016 |
Shubham Ajmera <shubhamajmera@google.com> |
Add unit tests for java.util.Calendar
and GregorianCalendar
Tests added for:
Calendar.getWeekYear()
Calendar.getWeeksInWeekYear()
Calendar.isWeekDateSupported()
Calendar.setWeekDate(int, int, int)
GregorianCalendar.getWeekYear()
GregorianCalendar.getWeeksInWeekYear()
GregorianCalendar.isWeekDateSupported()
GregorianCalendar.setWeekDate(int, int, int)
Bug: 27330986
Change-Id: I24faf062ec966c097ba84a09292654d1e178e9a4
|
| 90a3721dc9454a67e1c64409782d277424d38fc3 |
|
14-Jan-2016 |
Paul Duffin <paulduffin@google.com> |
Fix setting a calendar to a time during transition into DST
The transition into Daylight Savings Time (DST), i.e. spring
forward, results in a period of time (equal to the DST offset)
during which the wall clock time is incorrect. Basically, the
GregorianCalendar does not handle that situation properly and
calculates the wrong UTC time (in millis since 1st January
1970).
The reason this exists in Enso is because in OpenJDK the
handling of that situation is not performed directly in
GregorianCalendar computeTime() but instead in a special method
in sun.util.calendar.ZoneInfo. That means OpenJDK has similar
behavior to enso when a different TimeZone implementation, such
as SimpleTimeZone is used. That special code was removed so it
fell back to the code that is used for SimpleTimeZone which is
broken.
In pre-enso GregorianCalendar has a special method getOffset()
that is called for all instances of TimeZone that handles the
case so unlike OpenJDK it does work for SimpleTimeZone.
The solution is to add some code into GregorianCalendar that
handles the transition into DST specially. That means it will
work the same as pre-enso, not OpenJDK for SimpleTimeZone. As
it is complex I have refactored the code a bit more than
necessary in order to improve the structure and add some extra
documentation. I also added extra tests, some of which do not
work in OpenJDK.
Bug: 25897679
Change-Id: I06d0c81e8da20ec00267ed8ba24b26a8a2c163ed
|
| dda702b5a424ec84623e1a1e285c3ba0d515c131 |
|
17-Dec-2015 |
Paul Duffin <paulduffin@google.com> |
Fix, expand and move test for GregorianCalendar change date
The part of the test_getLeastMaximumI test that is broken is
checking a fix that was added in
https://issues.apache.org/jira/browse/HARMONY-2947. The fix
aimed to make the method obey its contract and take into
consideration the current values of the getFirstDayOfWeek,
getMinimalDaysInFirstWeek, getGregorianChange and
getTimeZone methods but it didn't.
The test broken because OpenJDK considered TimeZone, and the
test used the default time zone so it varied from run to run
which explains why it worked in some cases but not in others.
There are a number of other fields (apart from WEEK_OF_YEAR that
are affected by the transition between Gregorian and Julian
calendars so the part of the test that was broken was moved to
libcore.java.util.GregorianCalendarTest, fixed to work properly
by setting the TimeZone explicitly and expanded to test the
other fields too.
Bug: 25998114
Change-Id: I4fb6ed6143c682fce6cd2029d09bd59b17b3541d
|
| 3adfa669da17e644700961647574b2cb7a309139 |
|
17-Dec-2015 |
Paul Duffin <paulduffin@google.com> |
Fix test to match behavior of OpenJDK and RI
Bug: 26023857
Change-Id: Ib0632d9b04e54d38dc0e7bff18b6772ea605fc79
|
| 4eb6dd5362871b00084b2fe0cfb4f4b87f052d4c |
|
20-Nov-2013 |
Narayan Kamath <narayan@google.com> |
Document some surprising GregorianCalendar behavior.
bug: https://code.google.com/p/android/issues/detail?id=61993
Change-Id: I1725808187f23ddaa6ab1d14348c3b7702a5f833
|