Yongxing Wang wrote:
Emmanuel,
I am interested in the fix you made for MINA-678 where NioProcessor take
100% of the cpu. I assume you take the approach to ditch the selector
once detecting it is in epollWait() state and create a new one.
It's not working yet...
I realized that the fixed version is 2.0.0-RC1, I am wondering how big
the difference is between 2.0.0-RC1 and M6? Is it possible for you to
tell us the diff (hopefully, not a whole lot) so I can backport to M6 in
my local build and then tested it out?
There is a branch I created to experiment around this idea :
http://svn.apache.org/repos/asf/mina/branches/select-fix/
It contains all MINA, with the fix (work in progress).
IMO, as soon as this bug is fixed, it should be injected in the next
release, whatever it is (2.0.0-M7, M8 or RC1). RC1 will just be a
release candidate, ie the last versions before the GA. (It gives us the
opportunity to get some feedback from users for a few weeks instead of
releasing straightforward something which is not perfect)
Also, I guess the underlying jdk problem is not fixed in the JDK 7
Preview on the Sun website.....
It is supposed to be fixed on Java 7, b55, and will be backported in
java 6-014, due by december this year, AFAIK.
Thanks!!
Yong
--
--
cordialement, regards,
Emmanuel Lécharny
www.iktek.com
directory.apache.org
