[TIP] RFC: Post-assertRaises assertions WAS: ANN: unittest2 0.4.0 Release

[exarkun at twistedmatrix.com]

On 01:10 pm, olemis at gmail.com wrote:
>On Thu, Apr 8, 2010 at 5:42 PM, Michael Foord 
><michael at voidspace.org.uk> wrote:
>>Hello all,
>
>:o)
>>I'm pleased to announce the release of version 0.4.0 of unittest2.
>[...]
>>
>>For full details of all the new features, see my blog entry:
>>
>>* 
>>http://www.voidspace.org.uk/python/weblog/arch_d7_2010_04_03.shtml#e1171
>
>I wanted to know if you have considered support for writing assertions
>targeting the exception object been raised . In my experience that is
>very helpful when testing some kind of systems. For instance, in Trac
>XmlRpcPlugin (<=0.1.6, current trunk ;o) you can find many tests
>(actually all TCs checking for error conditions ;o) like this
>
>Therefore, we're recently considering [1]_ (not committed yet ;o) to
>adopt this style
>
>{{{
>#!python
>
>class TracRpcTestCase(unittest.TestCase):
>        # Further code omitted ;o)
>        def assertRaises(self, excClass, callableObj, *args, **kwargs):
>            """Enhanced assertions to detect exceptions."""
>            try:
>                callableObj(*args, **kwargs)
>            except excClass, e:
>                return e
>            except self.failureException :
>                raise
>            except Exception, e :
>                if hasattr(excClass, '__name__'): excName = 
>excClass.__name__
>                else: excName = str(excClass)
>
>                if hasattr(e, '__name__'): excMsg = e.__name__
>                else: excMsg = str(e)
>
>                raise self.failureException("\n\nExpected %s\n\nGot %s : 
>%s" % (
>                                        excName, e.__class__.__name__, 
>excMsg))
>            else:
>                if hasattr(excClass,'__name__'): excName = 
>excClass.__name__
>                else: excName = str(excClass)
>                raise self.failureException, "Expected %s\n\nNothing
>raised" % excName

Twisted Trial also includes an assertRaises like this.  It has worked 
well for many years in a wide variety of contexts.

Guido seemed to veto the API (<http://mail.python.org/pipermail/python- 
dev/2008-April/078726.html>):

  Hm. I've got to say that returning the exception object is, um, an odd
  API in the set of unittest APIs. I can see how it's sometimes more
  powerful, but I'd say that in many cases assertRaisesWithMessage will
  be easier to write and read. (And making it a regex match would be
  even cooler.)

IMHO, the (lack of) reasoning here is bogus.  In fact, even Guido seems 
to have forgotten that he doesn't like it 
(<http://bugs.python.org/issue2578#msg84622>):

  Antoine Pitrou <pitrou at free.fr> added the comment:
  > Wouldn't it be simpler to make assertRaises return the exception and 
let
  > the calling code match it as it feels like?

  Hm, that sounds awfully familiar. I can't recall if there was ever a
  good reason not to do this.

Jean-Paul