"Dmitrij D. Czarkoff" <czark...@gmail.com> writes: > Jérémie Courrèges-Anglas said: >> > https://github.com/gevent/gevent/issues/446 >> > >> > sigh >> >> So to work around this issue the following patch could be used. > [...] >> $OpenBSD$ >> --- dulwich/contrib/test_swift_smoke.py.orig Sun Mar 22 12:52:07 2015 >> +++ dulwich/contrib/test_swift_smoke.py Tue Apr 7 09:05:28 2015 >> @@ -41,6 +41,11 @@ import gevent >> from gevent import monkey >> monkey.patch_all() >> >> +# force the 'subprocess' module to use select() instead of poll(); >> +# the latter gets removed from the 'select' module by >> gevent.monkey.patch_all() >> +import subprocess >> +subprocess._has_poll = False >> + > > I would prefer to see gevent monkeypatching this. Actually, upstream is > waiting for such patch as said in discussion you've linked.
Almost one year, a problem created by the intrusive behavior of gevent, and yet no fix. Clearly they don't care. gevent-1.0.1 does not fix this issue and now bundles c-ares and libev. Yay. >> Additionally we could add py-mock as a test dep on devel/py-mock. > > Yes, that makes sense. Well, I'll commit that. >> What about running nosetests with the -v flag? > > That would make passing tests more verbose without changing anything > with respect to failing tests. I don't think it's useful. As you want, but if -v had been used earlier the missing dep on py-mock would have been obvious. -- jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF DDCC 0DFA 74AE 1524 E7EE
