JDK 10 RFR of JDK-8181395: Refactor several java/nio locale related shell tests to java

[Amy Lu]

On 6/7/17 2:14 PM, Amy Lu wrote:
> webrev: http://cr.openjdk.java.net/~amlu/8181395/webrev.00/
Hi, Alan

Is this fine to go?

Thanks,
Amy


On 6/9/17 12:46 AM, Amy Lu wrote:
> On 6/8/17 10:48 PM, Alan Bateman wrote:
>> On 07/06/2017 16:14, Amy Lu wrote:
>>> Defaul.java
>>> System.out.println(Charset.defaultCharset());
>>>
>>> The output for "nonexist" (not supported) LC_ALL is (at least on 
>>> Solaris): US-ASCII
>>>
>>> The output is the same as set LC_ALL from commandline:
>>>
>>> $ export LC_ALL=nonexist
>>> bash: warning: setlocale: LC_ALL: cannot change locale (nonexist)
>>>
>>> $ echo $LC_ALL
>>> nonexist
>>>
>>> $ jdk/bin/java Default
>>> US-ASCII
>>>
>>> I'm not sure why though. Also not sure whether it's the same across 
>>> platforms, settings. So I make it run with the "nonexist" locale to 
>>> get the String UNSUPPORTED.
>>>
>> I assume "US-ASCII" is coming from nl_langinfo(CODESET), at least on 
>> OSX/Linux/Solaris. I'm surprised that it prints "US-ASCII" on Windows 
>> but maybe that needs a system without the JP locales to prove that.
> This test won't run on Windows.
>
> DefaultCharsetTest.java
>     @requires (os.family == "linux" | os.family == "solaris")
> which is from the original default.sh
>
> Maybe this test should be expanded to OSX, I can file a separate bug 
> for this purpose (if needed).
>
> Thanks,
> Amy
>>
>> -Alan.
>