Stepan,
I 've filed it as HARMONY-1491. As far as I understand someone should
close it according to the latest Non-bug differences from RI policy.
Thanks,
2006/9/13, Stepan Mishura [EMAIL PROTECTED]:
On 9/13/06, Alexei Zakharov wrote:
Ok, Stepan, in this case I suggest to leave the check
Ok, Stepan, in this case I suggest to leave the check and rise the
additional Non-bug differences from RI JIRA (I can do if no one
objects). I don't really think there are many applications that rely
on this silent RI behavior, and IMHO we should not care until we
encounter one.
Regards,
On 9/13/06, Alexei Zakharov wrote:
Ok, Stepan, in this case I suggest to leave the check and rise the
additional Non-bug differences from RI JIRA (I can do if no one
objects). I don't really think there are many applications that rely
on this silent RI behavior, and IMHO we should not care
Hi Stepan,
Thank you for your attention to my patch first of all. IMHO everything
is ok except for the null-check you add to the setSource() method. It
seems RI does not check for null in this case. At least your
regression test fails on Sun JDK 1.5.0_06:
No expected NullPointerException
Alexei,
We have the following RI behaviour here:
1) Constructor doesn't allow 'null' value and throws NPE
2) setSource allow 'null' value
This looks inconsistent - to assign soure null value we can not use
constuctor directly!
Thanks,
Stepan.
On 9/12/06, Alexei Zakharov wrote:
Hi Stepan,
Just have found in java.beans package description:
Unless explicitly stated, null values or empty Strings are not valid
parameters for the methods in this package. You may expect to see exceptions
if these parameters are used.
So it is a bug in RI.
-Stepan.
On 9/12/06, Stepan Mishura wrote: