[threeten-dev] some problems in java.time.YearMonth and java.time.Year

[Stephen Colebourne]

Fixed in
http://hg.openjdk.java.net/threeten/threeten/jdk/rev/97e8eb496b37
http://hg.openjdk.java.net/threeten/threeten/jdk/rev/0e98b03f696a
http://hg.openjdk.java.net/threeten/threeten/jdk/rev/0dcbfca54160

Additional tests in
http://hg.openjdk.java.net/threeten/threeten/jdk/rev/72bb5d877b83

thanks
Stephen


On 12 March 2013 15:11, Stephen Colebourne <scolebourne at joda.org> wrote:
> On 9 March 2013 07:14, patrick zhang <patrick.zhang at oracle.com> wrote:
>> 1. Year.minus(TemporalAmout) does not work properly.
>> ===========
>>         Year year = Year.of(5);
>>         Year result = year.minus(Period.ofYears(2)) ;
>>         System.out.println(result);
>> ===========
>> Personally I think above code should work and return Year.of(3). But it
>> throws exception:
>
> This is a bug.
>
>> 2. javadoc for Year.minus(long, TemporalUnit) is wrong.
>> As you know the method will return Year same with minus(TemporalAmout) while
>> it is "YearMonth" in current javadoc:
>
> This is a bug.
>
> 3. javadoc for
>> java.time.Temporal.ChronoUnit.ERAS is confused.
>> Here there are 2 problems.
>> ====================
>> Unit that represents the concept of an era. The ISO calendar system doesn't
>> have eras thus it is impossible to add an era to a date or date-time. The
>> estimated duration of the era is artificially defined as 1,000,00,000 Years.
>> ====================
>> I guess it should be 1,000,000,000 years nor 1,000,00,000
>
> This is a bug.
>
>> ====================
>>         Year year = Year.of(-1);
>>         Year result = year.plus(1, ChronoUnit.ERAS);
>>
>>         System.out.println(year.get(ChronoField.ERA) + " " +
>> year.get(ChronoField.YEAR_OF_ERA) + " " + year.get(ChronoField.YEAR));
>>         System.out.println(result.get(ChronoField.ERA) + " " +
>> result.get(ChronoField.YEAR_OF_ERA) + " " + result.get(ChronoField.YEAR));
>>
>> output:
>> 0 2 -1
>> 1 2 2
>> ====================
>> Since ERAS is not supported in ISO calendar, I think it should not return
>> such a strange result.
>> It looks it only plus 1 on ChronoField.ERAS field simply......
>
> The ISO calendar system does support eras, see IsoEra. BCE is from
> proleptic year 0 backwards and CE is from proleptic year 1 forwards,
> just like AD/BC in GregorianCalendar.
>
> The code above is therefore correct, as it changes 2BCE to 2CE.
>
> Stephen