Open hoffmabc opened 10 years ago
This is because of wonky error handling expecting a reason from urllib2, which may not be returned. There need to be independent handlers for URLError and HTTPError, HTTPError returns no reason. URLError handler should be tolerant of no reason or even a reason filled with exploits and fuzz.
Not sure if this is related to #39 but I am seeing issues intermittently where a null exception is raised.