Hi, I was trying to write a blackbox test for a command in a plugin which requires custom rules for the test, and figured out that its not so easy to do that at the moment. this patch does the following: * extract rules.reset_rules() which instantiates our default user-based rule object which can be called from tests to reload it. * tree.iter_search_rules now uses the default parameter differently to ensure that if we change our rules._per_user_searcher after tree.py loaded, that the new one will get used. with it, its possible to write rule-dependant blackbox tests similar to some of the blackbox tests relying on configs. eg. def test_something(self): config.ensure_config_dir_exists() open(rules.rules_filename(),'wb').write(""" [name *.py] foo=bar """) rules.reset_rules() # run test that depends on the rule out,err = self.run_bzr('my-command') -- <| regards U| Marius H| <><