From 80f590288b1dc46f54718f6bfd71117c967911fa Mon Sep 17 00:00:00 2001 From: Ronny Pfannschmidt Date: Sun, 27 Jan 2013 02:10:29 +0100 Subject: [PATCH] add some bits to ISSUES --- ISSUES.txt | 54 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 54 insertions(+) diff --git a/ISSUES.txt b/ISSUES.txt index 446d5b9c9..8a69511fd 100644 --- a/ISSUES.txt +++ b/ISSUES.txt @@ -348,3 +348,57 @@ This would probably be done by marking:: ... With the possibility of doing this globally in an ini-file. + + +fnmatch for test names +---------------------- +tags: feature-wish + +various testsuites use suffixes instead of prefixes for test classes +also it lends itself to bdd style test names:: + + class UserBehaviour: + def anonymous_should_not_have_inbox(user): + ... + def registred_should_have_inbox(user): + .. + +using the following in pytest.ini:: + + [pytest] + python_classes = Test *Behaviour *Test + python_functions = test *_should_* + + +mechanism for running named parts of tests with different reporting behaviour +------------------------------------------------------------------------------ +tags: feature-wish-incomplete + +a few use-cases come to mind: + +* fail assertions and record that without stopping a complete test + + * this is in particular hepfull if a small bit of a test is known to fail/xfail:: + + def test_fun(): + with pytest.section('fdcheck, marks=pytest.mark.xfail_if(...)): + breaks_on_windows() + +* divide functional/acceptance tests into sections +* provide a different mechanism for generators, maybe something like:: + + def pytest_runtest_call(item) + if not generator: + ... + prepare_check = GeneratorCheckprepare() + + gen = item.obj(**fixtures) + for check in gen + id, call = prepare_check(check) + # bubble should only prevent exception propagation after a failure + # the whole test should still fail + # there might be need for a loer level api and taking custom markers into account + with pytest.section(id, bubble=False): + call() + +