[15] RFR: 8246662: Test java/time/test/java/time/format/TestUnicodeExtension.java failed on japanese locale.

[naoto.sato at oracle.com]

Hi Roger,

Thanks for the review.

On 6/9/20 8:52 AM, Roger Riggs wrote:
> Hi Naoto,
> 
> Since the default locale is being changed even briefly, the test should 
> be run in /othervm.
> Add:  @run testng/othervm ...

All java/time/test/java/time/format tests are run in othervm mode by 
default, defined in java/time/test/TEST.properties.

> 
> DateTimeFormatter:1498:
>     Can the optimization be retained in the case where z.equals(zone)?
>     Move it down to 1508+ and check the zone vs z.

It's not only zone, but also has to check chrono and decimalStyle. In 
the previous version, I intentionally did not keep the optimization, as 
it would no longer be "fast" since it would need to check those 
equality, but at least it would save creating the new DateTimeFormatter 
instance, so I resurrected it. Here is the updated webrev:

http://cr.openjdk.java.net/~naoto/8246662/webrev.01/

Naoto

> 
> Thanks, Roger
> 
> 
> 
> On 6/8/20 5:06 PM, naoto.sato at oracle.com wrote:
>> Hello,
>>
>> Please review the fix to the following issue:
>>
>> https://bugs.openjdk.java.net/browse/JDK-8246662
>>
>> The proposed changeset is located at:
>>
>> https://cr.openjdk.java.net/~naoto/8246662/webrev.00/
>>
>> This is a regression caused by the fix to JDK-8244245, where the 
>> locale related fields in the formatter are now overridden by 
>> localizedBy() method. It was only comparing the locale object for the 
>> fast path, but it was not sufficient.
>>
>> Naoto
>