Hi Rachna,
macosx//HostLocaleProviderAdapterImpl.java:
- line 172-177, might be a good place to use the new
ConcurrentMap.computeIfAbsent
- line 197: you could pre-size the StringBuilder since the target
length can be estimated
(unless they are all less than the default 16)
macosx && windows//HostLocaleProviderAdapterImpl.java
- toJavaTimeDateTimePattern() - is there a way to avoid having two
copies of this function?
Perhaps as a static method in JavaTimeDateTimePatternImpl.java.
The noreg-hard label on issue indicates testing is difficult and
specific to OS and host
configuration but it also seems unusual to have this much code and not
have a regression test.
If Masayoshi is satisfied and you have tested it in the target
configuration then
perhaps it is not worthwhile to invest in a special case regression test.
Thanks, Roger
On 11/30/2016 4:39 AM, Masayoshi Okutsu wrote:
Looks good to me.
Masayoshi
On 11/22/2016 6:30 PM, Rachna Goel wrote:
Hi,
Please review fix for JDK-8075577.
Bug : https://bugs.openjdk.java.net/browse/JDK-8075577
webrev : http://cr.openjdk.java.net/~rgoel/JDK-8075577/webrev.01/
Fix is to introduce new private spi
"sun.text.spi.JavaTimeDateTimePatternProvider.java" to retrieve
LocaleProvider specific Date/Time Patterns for "java.time" .
Thanks,
Rachna
