Re: [mevenide-user] Issues with maven-surefire-plugin 2.4 and system properties

Well,
it seems you test is forked but the system properties are copied over.
A result of http://jira.codehaus.org/browse/SUREFIRE-121 fix.
I'm going to reopen and add our case there..

Milos

On Jan 16, 2008 10:16 PM, Andreas Loew <[email protected]> wrote:
> Milos, All,
>
> it looks like the NetBeans Maven integration (for NB 5.5.1 as well as
> for 6.0) has a serious issue with the newly released
> maven-surefire-plugin 2.4 (released on Jan 15th, 2008).
>
> It looks like, although my tests are configured to use either
> forkMode=once or forkMode=perTest, Mevenide and/or the
> maven-surefire-plugin in 2.4 either do not fork or (most probably) start
> these new JVMs by explicitly (and incorrectly) passing all the system
> properties and property values of the currently running JVM (i.e. the
> JVM running the Netbeans IDE).
>
> This means these system properties have "interesting" values when the
> JUnit test code is running, like e.g.:
>
> javax.xml.parsers.DocumentBuilderFactory=org.netbeans.core.startup.DOMFactoryImpl
> javax.xml.parsers.SAXParserFactory=org.netbeans.core.startup.SAXFactoryImpl
>
> As these NetBeans factory classes are not available from my project's
> Meven test classpath, all my XML-related tests do now fail.
>
> In addition, I get a message about a missing Logger configuration class:
>
> Logging configuration class "org.netbeans.core.startup.TopLogging" failed
> java.lang.ClassNotFoundException: org.netbeans.core.startup.TopLogging
>
> The announcement for maven-surefire-plugin 2.4:
>
> http://www.nabble.com/-ANN--Maven-Surefire-Plugin-2.4-for-Maven-2-Released-tt14853145s177.html#a14853145
>
> says about this:
>
> "3) System properties [SUREFIRE-416]: Surefire allows you to configure
> the system properties of the currently running test.  By default,
> Surefire will run tests in a separate process (forkMode=once); if
> process forking is disabled (forkMode=never), Surefire will set the Java
> system properties of the currently executing Maven process, and finally
> change them back. This may have unanticipated side effects if you're
> running Maven in embedded mode and process forking is disabled.
> Embedders should not disable forking during Surefire testing.  [Tests
> themselves might do something even more dangerous, like calling
> System.exit().]"
>
> Switching back to the previous version of the surefire plugin (2.3.1)
> makes my tests work fine again:
>
> <plugin>
>    <groupId>org.apache.maven.plugins</groupId>
>    <artifactId>maven-surefire-plugin</artifactId>
>    <version>2.3.1</version>
> </plugin>
>
> Is this an issue with Mevenide or the new maven-surefire-plugin 2.4?
>
> Many thanks in advance for any insights!
>
> Best regards,
>
> Andreas
>
> --
> Andreas Loew
> Java Architect
> Sun Microsystems (Germany)
>
>
> ---------------------------------------------------------------------
> To unsubscribe from this list please visit:
>
>     http://xircles.codehaus.org/manage_email
>
>

---------------------------------------------------------------------
To unsubscribe from this list please visit:

    http://xircles.codehaus.org/manage_email