[
http://jira.codehaus.org/browse/MEXEC-14?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jerome Lacoste updated MEXEC-14:
--------------------------------
Fix Version/s: (was: before-1.1)
1.1-beta-1
> Don't have plugin's classloader on classpath for executed code
> --------------------------------------------------------------
>
> Key: MEXEC-14
> URL: http://jira.codehaus.org/browse/MEXEC-14
> Project: Maven 2.x Exec Plugin
> Issue Type: Improvement
> Affects Versions: before-1.1
> Reporter: David Smiley
> Assignee: Jerome Lacoste
> Priority: Minor
> Fix For: 1.1-beta-1
>
> Attachments: dsmiley_2785.diff
>
>
> The plugin was using a special ClassLoader (IsolatedClassLoader) which it
> defined, and the parent ClassLoader was set to that of the plugin (i.e.
> Maven's ClassLoader). I don't think that is desired, we want to pretend Java
> is being invoked from the command-line free of Maven (at least that's the
> illusion). We also want to do child-first delegation to be consistent with
> Java. So I swapped out IsolatedClassLoader with a simple application of
> URLClassLoader. Less code -- a good thing. Fixing this, however, revealed
> some bugs in the testing environment which relied on the former (wrong)
> behavior:
> * I needed to ensure that the test classpath of the plugin got onto the
> classpath of the pom being tested. It wasn't there before but should have
> been. This isn't strictly necessary but lets our tests remain easy by having
> test poms refer to a mainclass that's in exec plugin's test sources. That
> was relatively easy to do by on the fly modifying the outputDirectory
> configuration element to be that of the test classes output of exec plugin.
> That worked.
> * The test file ExecMojoArgsTest was broken and could not be easily fixed in
> place, if at all (it used MavenEmbedder). I re-wrote that test and
> incorporated the rewrite into ExecJavaMojoTest and then I deleted
> ExecMojoArgsTest (with permission from Philippe Jacot).
> In the attached patch, I also cleaned up lingering issues from MEXEC-6 a tad:
> * logging some things that aren't a problem at debug and not higher
> * remembered to destroy the ThreadGroup now
> * fixed bug where System properties should be stored prior to modification
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.codehaus.org/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
---------------------------------------------------------------------
To unsubscribe from this list, please visit:
http://xircles.codehaus.org/manage_email
