Mark Brouwer wrote:
Attached you will find a patch against the trunk for com.sun.jini.tool.ClassDep that contains the aggregated code changes for the issues RIVER-7, RIVER-8 and RIVER-151.
(I didn't try to run the code.) I'm a little confused about the lack of trailing file separator in [in,out]sideRoots. It looks like -inroot a.foo will cause matches with a.foobar.C for example, is that what you intended, or am I misreading the code? It seems desirable to add explicit checking and failure if the user supplies mixed new/old options. A few suggested wording changes: exluded -> excluded Any found classes that is in -> Any classes found in If not specified all packages are considered to be in the <code>inroot</code> namespace. -> If not specified all found classes are considered roots. Indicates whether the root directories specified must for finding classes to include for dependency checking in the 'old' way, or that the new behavior must be effective. -> Indicates whether the root directories specified for finding classes for dependency checking must be traversed in the 'old' way, or that the new behavior must be effective. reagard -> regard - Bob
