RFR: 8006039 : test/tools/launcher/I18NJarTest.java fails on Mac w/ LANG=C, LC_ALL=C

Kumar Srinivasan kumar.x.srinivasan at oracle.com
Mon Feb 25 15:10:58 UTC 2013

Hi Brent,

Ok I tested on Windows with JA installed and your patch works,
also tested on Solaris.

On other nit,   Copyright on I18NJarTest.java needs updating.

Looks good, go for it.



> On 2/21/13 7:33 PM, Kumar Srinivasan wrote:
>> A nit
>> - " LC_ALL" + LC_ALL + "\n" +
>> + " LC_ALL=" + LC_ALL + "\n" +
> Done, thanks.
>> If Mac is the only system affected by this, should we make the checks 
>> for
>> LC* and LANG specific to Macs since other platforms don't have this 
>> issue ?
>> I am concerned this might pass vacuously on systems configured 
>> correctly.
> LANG/LC_ALL=C is not a correct configuration for running this test.  
> It ought to pass vacuously in this case (and does, AFAIK, on other 
> platforms).
> What is Mac-specific is that sun.jnu.encoding is now always set to 
> UTF-8, even under LANG/LC_ALL=C.  The test sees this, and tries (and 
> fails) to run the "real" test instead of passing vacuously.
> I don't think it does any harm to check for LANG/LC_ALL != 'C' on 
> other platforms.  But it's only on Mac that it's required in order to 
> detect a mis-configured test environment.
>>> Of course it would be even better to update the test so the automated
>>> test machines actually test the intended functionality.
> >> ...
> >> I can file a separate bug for this.
> I've filed 8008761 (test/tools/launcher/I18NJarTest.java should 
> automatically setup its environment) to cover this.
> To be clear, my short-term goal is to take I18NJarTest back off the 
> ProblemList.  The test does correctly test the intended functionality 
> under the default Mac environment (LANG=en_US.UTF-8).  So I want to 
> get it running again for engineers using the "jdk_tools" test target 
> on their local Mac (without causing nightly test failure noise).
> Later, as time permits, I can tackle 8008761, and get the test to 
> configure its own environment.
> Thanks,
> -Brent

More information about the core-libs-dev mailing list