Wouter van Heyst added the comment: What the patch does is replace the hard-coded address 127.0.0.1 for the loopback interface with a one-time lookup. I'd argue the hunk @@ -315,7 +316,7 @@ should be dropped.
The two arguments that motivate this change are: 1) The address of the loopback interface is not always 127.0.0.1, binding to it may fail. 2) The tests seem to assume localhost resolves to the same address where the server is running. Working on this patch started with the ConnectionRefusedError that Håkan mentioned. Applying the patch made the ConnectionRefusedError go away, but I think we may have solved a different problem and suppressed the real problem as a side effect. @r.david.murray: would the patch be acceptable on dropping said hunk, and changing the description to "Do not assume localhost resolves to 127.0.0.1"? That still leaves, I think, a hidden bug somewhere in the proxy code that will surface in some corner case. ---------- _______________________________________ Python tracker <rep...@bugs.python.org> <http://bugs.python.org/issue22753> _______________________________________ _______________________________________________ Python-bugs-list mailing list Unsubscribe: https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
