Open UnitedMarsupials-zz opened 6 years ago
I am wondering if I should handle it in a single place (perhaps with AddrInfo
structure). I will commit fix in the next couple of days. Thanks again.
As mentioned above, I applied the fix in AddrInfo
structure so it actually simplifies all uses. I will apply similar patch for pycos as well. I only tested with OS X. Please let me know if this works for you.
I have committed changes to pycos as well. Please let me know if both packages work for you. You can use examples in pycos to test; e.g,. run dispycosnode.py
on a node and run dispycos_*.py
programs, possibly with different combinations of network setup, such as ipv4, ipv4_udp_multicast option, ipv6 etc.
Apple's Darwin is just one of the many BSD-derived systems, which are all particular about certain networking things. Wherever you are testing for "darwin", you should instead be testing for all of these.
For example, for
dispynode.py
to come up properly on FreeBSD, the following patch is necessary:Because the same check is needed in multiple places (dispy, dispynode, dispyrelay, dispyscheduler), under both
py2/
andpy3/
, it may be best to hide the list under a helper-function. Oh, and pycos has the same problem in its own twonetpycos.py
.Not sure, how to best fix it...
I would also recommend the following hunk: