Alexei Zakharov wrote:
Hi Gregory,
Thanks, as far as I can tell your patch fixes the reproducer and
doesn't add obvious regression at least. However, it seems while
testing your patch I've found another (minor) problem with JNI
implementation in DRLVM. I've filed it as HARMONY-4846 (" [drlvm][jni]
CallVoidMethod crashes VM if applied to Non-void method"). Since it is
about VM crash it may also deserve VM gurus' attention.
Thank you, I committed patches to fix both bugs. The 2nd one is arguable
since JNI spec states that programmer who writes native code is
responsible that it is correct. But fixing it is very easy, so I
committed a fix for it too.
2007/9/24, Gregory Shimansky <[EMAIL PROTECTED]>:
Alexei Zakharov wrote:
Hi all,
2) org.apache.harmony.beans.tests.java.beans.beancontext.BeanContextSupportTest
This passes ok for me.
I've just fixed this failure. Looks like it was introduced by commit
r578410 - most likely the patch haven't been checked on DRLVM. The
problem itself was in differences between JNI implementations in DRLVM
and J9/RI. For more details please see HARMONY-4840 and HARMONY-4841.
Alexei, thank you for finding this bug.
I created a patch for HARMONY-4841 that fixes CallVoidMethod
implementation in DRLVM when calling constructors. Would anyone mind if
I commit it?
My patch also fixed
org.apache.harmony.luni.tests.java.io.ObjectInputStreamTest.test_readObject_Hierarchy
failure reported at [1]. Therefore, all HUT tests should now pass on DRLVM.
[1] http://people.apache.org/~varlax/harmony-integrity/
Thanks,
Alexei
2007/9/17, Tim Ellison <[EMAIL PROTECTED]>:
Alexei Zakharov wrote:
I'm able to reproduce and currently investigating the following HUT
failures reported in [1]:
Thanks Alexei. If it helps track down the problems, when I run all the
classlib tests on Win ia32 with the IBM VME...
(in that order):
1) java.awt.EventDispatchThreadRTest
this also fails for me
2) org.apache.harmony.beans.tests.java.beans.beancontext.BeanContextSupportTest
This passes ok for me.
--
Gregory
