[
https://jira.codehaus.org/browse/SUREFIRE-1122?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=358894#comment-358894
]
Andreas Gudian commented on SUREFIRE-1122:
------------------------------------------
Got it. Fixed it. The StatelessXmlReporter wasn't that stateless anymore, but
ogt re-created for each new thread and each test class execution in the
concurrent run listener. The flaky-rerun acts like rerunning the test class
from the outside, so the StatelessXmlReporter got instantiated again and forgot
what it knew about the previous run of that class. The map containing that
information was now pulled out of that reporter to make it stateless again
(that's easier than renaming the class, right? ;) ).
I'll commit it tomorrow after creating a nice IT out of Roys test case... :)
> When running failsafe with -Dfailsafe.rerunFailingTestsCount and parallel
> all, the reports are not generated correctly
> ----------------------------------------------------------------------------------------------------------------------
>
> Key: SUREFIRE-1122
> URL: https://jira.codehaus.org/browse/SUREFIRE-1122
> Project: Maven Surefire
> Issue Type: Bug
> Components: Maven Failsafe Plugin, Maven Surefire Plugin
> Affects Versions: 2.18
> Reporter: Roy Arnon
> Assignee: Andreas Gudian
> Fix For: 2.18.1
>
> Attachments: invalid.xml, test.zip, valid.xml
>
>
> Hi,
> I just tried the new feature -Dfailsafe.rerunFailingTestsCount on our
> failsafe build.
> Unfortunately, when running the build using parallel=all, it seems the junit
> xml files are not generated correctly.
> I've tried running it multiple different ways, including classedAndMethods,
> and all, but it seems to generate an incorrect xml file either way.
> I've attached both files here.
> I can provide a pom and java test case if required.
--
This message was sent by Atlassian JIRA
(v6.1.6#6162)
