On 12/20/12 7:33 AM, Andrew Hughes wrote:
I was surprised to see these being backported. I can follow the reasoning for
doing so, but we must be careful of introducing unwanted behaviour changes
through a fix which isn't really required in a maintenance release.
I'm not thinking so much of the <> operator but this change also moves to use
try-with-resources. While the new behaviour may be more correct, it's still
may be a change from what was there before.
Hi Andrew,
This is a fair point. You're right to point out the behavior changes (such as
with try-with-resources), which are unusual for warnings fixes. In general when
working on or reviewing warnings changes I try to avoid any behavior changes. I
think this a good policy to stick to, but in this case I had made an exception.
Sorry you were surprised by it.
Also, this really is about maintenance of RMI. We are putting some effort into
making some specific, critical bugfixes in RMI, and so there will be behavior
changes. In this context it seems sensible to have the RMI source code and
behavior be identical between 7u and 8. It should make long term maintenance
easier.
You'll also note that we're putting effort into fixing up the RMI tests in 8
and backporting these changes to 7u.
As a heads-up, I have one more warnings fix for RMI that I'm preparing to
backport. This contains changes to the java.rmi packages, but also mixed in
with changes to other packages. I'm not actually sure of the best way to
proceed with this one. I'll post a patch at some point soon, and I'd encourage
you to look for it and comment on it.
s'marks
