Hi Brian, I have some concerns with this patch.
First, I think the stripTrailingZeros specification should explicitly call out what happens with a numerically zero value.
Second, I would have expected the code change to be limited to the stripTrailingZeros method. The private createAndStripZerosToMatchScale methods are used during rounding operations and unconditionally set the scale of a zero value to zero, which is incorrect according to the specification.
I believe sufficiently thorough tests of divide should fail with the current changes to createAndStripZerosToMatchScale.
Cheers, -Joe On 07/03/2013 08:46 AM, Brian Burkhalter wrote:
Hello, The patch here http://cr.openjdk.java.net/~bpb/6480539/ is proposed to resolve this issue http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=6480539 which was most recently discussed in this thread http://mail.openjdk.java.net/pipermail/core-libs-dev/2013-February/014252.html The proposed fix in this prior thread is withdrawn. Thanks, Brian
