Looks like the random failures are not limited to window. The travis
build (ubuntu), just failed with 44 failures, e.g.
AbstractMultiValuedMapTest$TestMultiValuedMapAsMap>AbstractMapTest.testMapToString:745->AbstractMapTest.verify:1947->AbstractMapTest.verifyMap:1958
hashCodes should be the same expected:<205172737> but was:<9372706>
see: https://travis-ci.org/apache/commons-collections/jobs/282169803
Am 28.09.2017 um 20:23 schrieb Andreas Kuhtz:
Hi,
2017-09-27 22:28 GMT+02:00 Bruno P. Kinoshita <ki...@apache.org>:
Forgot to mention something important. When I was debugging the hash
tests, they would pretty much always pass.
But if I ran the tests normally in maven or eclipse they would always fail.
Debugging adds a small overhead I think, that could be slowing down some
part of the code with a concurrency issue?
That's an explanation. If I start a Virtualbox instance that prepares some
load I can see the tests pass under Java 9 ... I had the Virtualbox running
yesterday when the tests passed.
Just in case it helps...
Interesting this surefire configuration Andreas. Do you have more
information about it? Is it something we may have to worry about in
components that use locale and jvm 9?
I ran into an issue with the language selection dialog of izpack and found
that this configuration solved the problem there.
It's noted in the release notes of Java 9:
http://www.oracle.com/technetwork/java/javase/9-relnote-issues-3704069.html#JDK-8008577
Cheers,
Andreas
Cheers
Bruno
Sent from Yahoo Mail on Android
<https://overview.mail.yahoo.com/mobile/?.src=Android>
On Thu, 28 Sep 2017 at 4:45, Andreas Kuhtz
<andreas.ku...@gmail.com> wrote:
Hi,
I'm running Win7 and with "mvn clean verify". Checkout master.
I got the hash code errors, too. But sometimes the tests pass, the next
time they failed with 44 failures (the hash code verification).
As I'm not sure if my antivirus kicks in and keep some files in the target
dir, I ran "mvn clean" and after that "mvn clean verify".
But now I have the 44 failures with jdk-1.7.0_67, jdk-1.8.0_131 and 9+181
Gary:
To get around the issue with ServiceConfiguration sun.util.locale... I
added the following profile to the pom.xml. After that I only got the 44
errors (hash code) as with Java 7 and 8.
<profile>
<id>jdk9-build</id>
<activation>
<jdk>9</jdk>
</activation>
<build>
<plugins>
<plugin>
<artifactId>maven-surefire-plugin</artifactId>
<configuration>
<argLine>-Djava.locale.providers=COMPAT</argLine>
</configuration>
</plugin>
</plugins>
</build>
</profile>
I'm not sure if I can help more, but if you need more info just let me
know.
Andreas
2017-09-27 19:10 GMT+02:00 Gary Gregory <garydgreg...@gmail.com>:
Using git master and 'mvn clean verify', on Windows 10, with Java 7, I
get
44 failures, and I think this is all/mostly hash code problems.
With Java 8, all tests pass.
With Java 9, I get ONE failure: ListIteratorWrapperTest.testRemove:116 ╗
ServiceConfiguration sun.util.locale....
If I checkout the 4.1 tag and use Java 7, I get 44 failures. ARG! What
the
heck?
Gary
On Mon, Sep 25, 2017 at 12:02 PM, Benedikt Ritter <brit...@apache.org>
wrote:
Hi,
who can help with the test failures on Windows? I’d like to push out
the
next release soon to get Collections Java 9 ready. I don’t have access
to a
Windows installation and I don’t feel like debugging inside a VM :o)
Regards,
Benedikt
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org
For additional commands, e-mail: dev-h...@commons.apache.org
