>>>>> "CM" == Conor MacNeill <[EMAIL PROTECTED]> writes:
CM> I already modified it. If you specify a classpath and fork="no" , CM> the java task will use the classloader to load the class with the CM> given classpath. Do you think it needs more explit control than CM> that ? I was busy putting together my stuff - I had built it last night and took your commit as a signal to return to "business as usual". Therefore I didn't look at the actual code you had submitted but only read the description, sorry. My fear was that we'd use a classloader even if the user didn't specify a classpath of her own - now that I see that this is not the case I don't have any problems with it (why would you specify a classpath if not to use it). Maybe we should add an additional option to get a custom classloader even if the user didn't specify a classpath? This could be used to force the VM into reloading classes that might have changed since Ant has been started? How about integrating the classloader logic into ExecuteJava somehow and use that inside the Java task if fork was false? That way the same logic will lend itself to other <java> like tasks easily. Stefan
