* Improve the error handling code
Rather than multiple return paths, have a single return and set the message based
on the type of failure.
* Add another test for non-specific failures
* Reorganize tests so failure tests are in one tasks file
* Remove os.stat() call and add changelog
