Maybe you will take a look at http://enitsys.sourceforge.net/ant-classloadertask/
a little older, but works fine. (classloaderreport is broken with ant 1.8.x) Am 10.06.2012 11:54, schrieb Peter West:
I'm looking at a complex build that builds a jar file with a collection of classes, then runs a series of tests on the distribution that will include that jar. The tests affectively mirror what users of the distribution will be doing. When I do not have the generated jar in the startup classpath for the build, I can generate the jar file, define the tasks in terms of the jar file's classes, and begin to execute particular taskdef'd tasks. All good, and the classes are discovered. However, within the executing task, other classes from the jar are required. The attempt to load the first of these classes fails. It seems that within the executing task, even though the classpath for the execution of that task requires the new jar file, that class environment is lost. Hence the requirement for the initial classpath entry. Or so I believe. Is this correct? What is happening within the executing task that resets the classpath environment? Is there a way around this, such that I can define the classpath within the build? One possibility is to fork ant via a java task for the testing phase, but I am looking for other alternatives. Is antlib relevant to this problem? Peter West "Are not all these who are speaking Galileans?" --------------------------------------------------------------------- To unsubscribe, e-mail: user-unsubscr...@ant.apache.org For additional commands, e-mail: user-h...@ant.apache.org
--------------------------------------------------------------------- To unsubscribe, e-mail: user-unsubscr...@ant.apache.org For additional commands, e-mail: user-h...@ant.apache.org
