[TIP] pytest, multiprocessing and pyzmq's ioloop

Pere Martir pere.martir4 at gmail.com
Wed Oct 26 14:17:21 PDT 2011


On Wed, Oct 26, 2011 at 10:42 AM, Stefan Scherfke
<stefan.scherfke at offis.de> wrote:
> Holger, Pere,
> I just wrapped the zmq socket with another class. Its recv-method tries to
> call zmq's recv several times in non-blocking-mode and raises an exception
> after a while. That causes py.test to stop and shows me, which recv-call
> caused the error.

What a coincidence. I didn't suggest this method but I actually did
the same. However, it doesn't seem to be a good design pattern to me:
it's too far from the real use case - too many 'wrapper' code just to
make the test work. I'd like to know if there is some well known
anti-pattern.



More information about the testing-in-python mailing list