chatexchage/client.py: catch HTTPError for refused actions

[tripleee]

With this additional exception handler, the queue is properly drained when something fails, instead of retrying forever

(Closes: #164)

[makyen]

While removing the request from the queue upon an unresolvable HTTP error is probably a reasonable idea, this effectively causes all HTTP errors to be ignored without any way for the calling program to know that the error happened and/or that the requested action didn't take place. It would be quite a bit better to have some methodology to inform the calling program that the error occurred and that the specific request was dropped, rather than completed. Alternately, inform the calling program what error occurred, and the response from SE, and allow the requesting program the ability to drop the request or to retry the request.

[Manishearth]

Hmm, yeah. I don't have the time to design such a thing but would support it being added.

[tripleee]

The threading model makes it hard to get back to where you were before you started a background thread to take care of the queue. I'm not saying it can't be done, but it's more complicated than just observing the failure and reporting back to the caller.

The problem in #164 was precisely that the error was lost forever, but the symptom was also that the program would hang around until you killed it. This actually produces a traceback in my limited testing, though there isn't much you can do from the place where you originally tried to post something.

Adapted from the earlier bug report, here is a simple demo program which now surfaces a traceback.

from time import sleep
from chatexchange.client import Client

client = Client('stackexchange.com', USER, PASS)
room = client.get_room(111121)
room.send_message('Testing')
sleep(15)
client.logout()