users@glassfish.java.net

Re: Absent Code attribute in method that is not native or abstract!

From: Sahoo <Sahoo_at_Sun.COM>
Date: Tue, 17 Jul 2007 20:55:31 +0530

Sorry to know that you had to waste so much of time for this issue. I am
trying to understand how we can improve documentation of our maven
artifacts. How did you come to know about this maven artifact? It has
intentionally got implementation-less classes. That blog very clearly
mentions this fact and says that it must only be used with *provider*
scope. IMO, we *should* have put this information in the pom.xml. Do you
read the comments in pom.xml before using a maven artifact?

Thanks,
Sahoo
glassfish_at_javadesktop.org wrote:
> Here is the answer
>
> http://weblogs.java.net/blog/ludo/archive/2007/01/java_ee_5_apis.html
>
> I was using the maven artifacts which are seriously stubbed jars. They only let you compile, not link runtime.
>
> This causes great problems when trying to use, e.g. EasyMock to mock classes for unit tests.
>
> The JDK 1.5 vs 1.6 issue is because 1.6 includes the persistence stuff itself. When forking, though, the forked classloader puts the stubbed versions first in the classpath, thereby breaking everything again.
>
> Three days wasted all because of somebody putting stubbed borked jars in a Maven repository. Only working jars should be in a maven repository!
> [Message sent by forum member 'stephenconnolly' (stephenconnolly)]
>
> http://forums.java.net/jive/thread.jspa?messageID=226931
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe_at_glassfish.dev.java.net
> For additional commands, e-mail: users-help_at_glassfish.dev.java.net
>
>