RE: [classlib][drlvm] classlib test tests.api.java.util.BitSetTest failed today on all platforms

[Ivaschenko, Aleksei V]

Hello,

I found that Jetty scenario recently started to fail because of bug in
java.util.BitSet. This simple test behaves differently on Harmony and
RI:

import java.util.BitSet;

public class BSTest {

    public static void main(String[] args) throws Exception {
        BitSet bs1 = new BitSet(256);
        BitSet bs2 = new BitSet(256);
        bs1.set('a');
        bs2.set('%');
        bs1.andNot(bs2);
        System.out.println("'a' allowed in bs1: " + bs1.get('a'));
    }
}

Harmony prints:

'a' allowed in bs1: false

while RI prints:

'a' allowed in bs1: true

I've checked it on r547680 - the problem still exists.

Tim, would you take a look please?

Thanks,
Aleksei.

>-----Original Message-----
>From: Vladimir Ivanov [mailto:[EMAIL PROTECTED]
>Sent: Friday, June 15, 2007 7:33 PM
>
>On 6/15/07, Tim Ellison <[EMAIL PROTECTED]> wrote:
>> Tim Ellison wrote:
>> > Vladimir Ivanov wrote:
>> >> Hello,
>> >> the classlib test tests.api.java.util.BitSetTest failed for me
today
>> >> on all platforms on DRLVM with log:
>> >>
>> >> Test: test_ConstructorIClass: tests.api.java.util.BitSetTest
>> >> java.lang.OutOfMemoryError
>> >>     <no stack trace available>
>> >>
>> >> Could someone reproduce/ take care about it?
>> >
>> > I'll take a look
>>
>> Creating a "new BitSet(Integer.MAX_VALUE)" works ok using the IBM
VME,
>> but causes an OutOfMemoryError running the same code on DRLVM.  The
RI
>> also fails with OOME, so I modified the test to ignore OOMEs.
>>
>> The test was introduced for a regression case that threw
>> NegativeArraySizeException, so the modification should be ok.
>>
>
>> Please check with code >= r547589.
>Thanks, it works now.
>
> thanks, Vladimir
>
>>
>> Regards,
>> Tim
>>