Regression introduced with JACC EJBMethodPermission

[anil.saldhana]

p0 = new EJBMethodPermission("someEJB", "methodX,,");
p1 = new EJBMethodPermission("someEJB", "methodX");
assertTrue("! p0.implies(p1)", p0.implies(p1) == false);

I may have introduced a regression with my fix to
http://jira.jboss.com/jira/browse/JBAS-5014

as evidenced by:
http://jira.jboss.com/jira/browse/JBAS-5057

The ctr and implies method of this permission is described here:
http://java.sun.com/j2ee/1.4/docs/api/javax/security/jacc/EJBMethodPermission.html

I need to verify whether this in fact is a regression.

Scott, thoughts?

[anil.saldhana]

I think it is a regression because p0 means a method with zero arguments whereas p1 means a perm for a specific method name but various overloaded arguments(all overloaded methods).

So, I think p1.implies(p0) == true where as p0.implies(p1) == false.

[anil.saldhana]

I need to rollback the change I made in JBAS-5014. I already did it for Branch_4_2. I need to do the same for javaee project.

UPDATE: Rolledback in javaee project also.

[starksm64]

I think it is a regression because p0 means a method with zero arguments whereas p1 means a perm for a specific method name but various overloaded arguments(all overloaded methods).

So, I think p1.implies(p0) == true where as p0.implies(p1) == false.

Yes, that is how these should evaluate.

[anil.saldhana]

The reason the JACC tests were failing were due to:

for(MethodPermissionMetaData perm : perms)
 {
 MethodsMetaData methods = perm.getMethods();
 if(methods != null)
 for(org.jboss.metadata.ejb.spec.MethodMetaData mmd : methods)
 {
 String[] params = {};
 if(mmd.getMethodParams() != null)
 params = mmd.getMethodParams().toArray(params);
 else
 params = null; <= I ADDED THIS NULL

Because the an empty array implies method with zero arguments, whereas a null array means all methods.