Looks good to me!
-- daniel
On 10/23/13 6:13 PM, Jaroslav Bachorik wrote:
On 23.10.2013 17:49, Daniel Fuchs wrote:
On 10/23/13 5:35 PM, Jaroslav Bachorik wrote:
On 23.10.2013 17:32, Daniel Fuchs wrote:
Hi Jaroslav,
Have you considered setting a limit on the number of tries you
will attempt, rather than just 'while (true)'?
It might avoid creating a busy while loop in case there's some global
networking issue.
It might be difficult to come up with the right number. Probably a few
hundred could be enough even for cases when you hit a contiguous range
of ports occupied by different applications.
Instead of using an arbitrary 12424 as the default starting point
it could be interesting to try with the first available port + 1;
something like:
final int startPort;
try (ServerSocket s = new ServerSocket(0)) {
startPort = s.getLocalPort() + 1;
}
System.out.println("startPort: "+startPort);
But maybe that's what the test that calls the code is already doing?
Nope, it's hardcoded.
I've added the cap for the number of consecutive checks for a port being
available and dynamic lookup of the first available port (if it's not
specified by the system property "rmi.port").
http://cr.openjdk.java.net/~jbachorik/8027058/webrev.01
Thanks for the additional ideas!
-JB-
-- daniel
-JB-
-- daniel
On 10/23/13 5:16 PM, Jaroslav Bachorik wrote:
Please, review the following test change:
Issue: https://bugs.openjdk.java.net/browse/JDK-8027058
Webrev: http://cr.openjdk.java.net/~jbachorik/8027058/webrev.00
The test fails when the requested port is unavailable. The test logic
already runs the subtests with different port numbers (increasing the
port number for each subtest) and this patch adds the logic to
increase
the port number in case of the subtest failure caused by
BindException.
This way the test is able to "autodiscover" free ports and run the
bootstrap subtests successfully.
Thanks,
-JB-