[rfc][icedtea-web] AccessClassInPackage Reproducer Failure

[Jiri Vanek]

On 11/13/2014 09:51 AM, Jiri Vanek wrote:
> On 11/12/2014 10:25 PM, Jie Kang wrote:
>> Hello,
>>
>>
>> I have been running a lot of reproducers lately and some consistent failures caught my eye today.
>> In the AccessClassInPackage reproducers, a number of asserts are made including the following two:
>>
>> Assert.assertFalse("AccessClassInPackageTestLunch1 should not be terminated, but was",
>> pr.wasTerminated);
>> Assert.assertEquals((Integer) 0, pr.returnValue);
>>
>>
>> The tests all fail due to these two asserts (wasTerminated is always true, not false, and return
>> value is always 130 (exit due to control-C? [1]), not 0). When removing these lines and checking
>> the tests, they all work and pass. The Class.forName call succeeds and is printed to stdout.
>>
>> So I am wondering, what are these two lines for? Is there some regression causing these failures?
>>
>>
>> P.S. patch attached is for easy reference to code I am talking about above; not an actual patch
>> (though we may want to apply a similar patch)
>>
>> [1] http://tldp.org/LDP/abs/html/exitcodes.html
>>
>>
>> Regards,
>>
>>
> hi!
>
> I will look into it later, but general hint is -  those are javaws tests. Javaws  should print it
> out and clsoe itself. Tahts the "Assert.assertEquals((Integer) 0, pr.returnValue);"  Fact that it do
> not close itself and is timouted - and so killed - is wrong. Probablyjavaws hangs.
>
>

So this really is serious regression. Our apps in head are keeping hanging much moire then in case 
of 1.5.


J.