would you please review the following patch?
bug: https://bugs.openjdk.java.net/browse/JDK-8188897
webrev: http://cr.openjdk.java.net/~mli/8188897/webrev.00/
Thank you
-Hamlin
Hi Hamlin,
Instead of a simple time delay, it would be useful to wait for the
RegistryVM to terminate.
In killRegistry: 149, adding subreg.waitFor() should be sufficient.
58: If using a 'for' loop it would be easier to understand if it
included the usual start, increment and termination.
In
Hello,
Some general comments on the coding for tests like this:
* It is preferable to avoid sleep in tests to avoid increasing the
minimum amount of time a test takes to run. This helps limit the overall
time it takes the test suite to run.
* If timeouts are used, it is recommend to factor t
Hi Joe, Roger,
Thank you for reviewing, I have refactored the test more and fix as you
suggested.
please review http://cr.openjdk.java.net/~mli/8188897/webrev.01/
Thank you
-Hamlin
On 04/04/2018 10:42 AM, Joseph D. Darcy wrote:
Hello,
Some general comments on the coding for tests like th
Hi Hamlin,
Reexport.java:
I think this change to use a separate process for the 2nd registry
changes the test so that it does not
address the original test case. The original problem was the incorrect
retention of an object in
the object table when the create of a registry in the same proces
Hi Roger,
I have changed to not use RegistryVM at all, please review the patch:
http://cr.openjdk.java.net/~mli/8188897/webrev.02/
Thank you
-Hamlin
On 04/04/2018 10:15 PM, Roger Riggs wrote:
Hi Hamlin,
Reexport.java:
I think this change to use a separate process for the 2nd registry
ch
Hi Hamlin,
Great, the new busy port algorithm looks better.
The port assigned will always be one that was available and is now busy
to prevent the registry creation.
As expected, there is a small window between the (try-with-finally)
close of the server socket channel
and the 2nd createReg.
On 10/04/2018 3:09 AM, Roger Riggs wrote:
Hi Hamlin,
Hi Roger,
Thank you for review.
Great, the new busy port algorithm looks better.
The port assigned will always be one that was available and is now
busy to prevent the registry creation.
As expected, there is a small window between the
Hi Hamlin,
Looks good,
Thanks for the updates.
On 4/9/2018 10:49 PM, Hamlin Li wrote:
On 10/04/2018 3:09 AM, Roger Riggs wrote:
Hi Hamlin,
Hi Roger,
Thank you for review.
Great, the new busy port algorithm looks better.
The port assigned will always be one that was available and is now