Made websocket subprotocols conform to spec#181
Conversation
|
Hmm. |
|
why do you think a warning is unsuited here? something wants to connect with a protocol that isn’t supported: that’s not normal, and a log entry might help debugging instead of silently succeeding with a return value that might not be checked. |
I see nothing wrong with my code if client requests non-supported protocol. Server code should not changed to "fix" warning, problem is inside incompatible client. |
|
I'm agree with @asvetlov In this case, I strongly vote to use a logger with warning level. If you use warnings instead of logging for that, I don't receive e-mail that something is wrong with a client. |
|
it’s up to him then to introduce logging, and this will wait? |
@flying-sheep sorry, I don't follow. |
|
I meant: this project doesn’t use any logging yet AFAICT. So you should introduce logging how you see fit (central logger with just basicConfig() or per-module logger, maybe even finer grained. using name, hardcoded names or custom logger classes, yadda yadda) and after i see how you want logging to be done, i’ll update this pull request to use logging. |
|
Do you talk about this: https://github.com/KeepSafe/aiohttp/blob/master/aiohttp/log.py ? |
|
Yes, I think ./aiohttp/log.py is good place for new logger. You can call it "aiohttp.websocket" for example. |
|
it appears that i am blind or looked at an old version or something. in any case, i didn’t know of the existence of |
0edbf5d to
98a5e16
Compare
98a5e16 to
8f881a7
Compare
|
done! |
Made websocket subprotocols conform to spec
|
Thanks! |
fixes #178