-
Notifications
You must be signed in to change notification settings - Fork 55
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Miscellaneous tweaks to transport handling and auth tests #296
Commits on Jun 27, 2016
-
Configuration menu - View commit details
-
Copy full SHA for 2081939 - Browse repository at this point
Copy the full SHA 2081939View commit details -
Configuration menu - View commit details
-
Copy full SHA for 4c388a2 - Browse repository at this point
Copy the full SHA 4c388a2View commit details -
Upgrade: stop preference timeout unconditionally on tryConnect cb
The original idea behind not stopping it was in case something stalled the websocket between 'viable' and 'connected' stages, so that the fallback would be used then. In practice we have no evidence that that's ever happened (an established websocket generally means we can use the websocket transport), and keeping the timeout running causes problems if the connection is successful but then something else disconnects it just before the timeout expires (eg a quick token expiry).
Configuration menu - View commit details
-
Copy full SHA for c643c0b - Browse repository at this point
Copy the full SHA c643c0bView commit details -
Configuration menu - View commit details
-
Copy full SHA for 3782b4e - Browse repository at this point
Copy the full SHA 3782b4eView commit details -
Use safeArrForEach for when the function might mutate the array
Previously, only every other transport in eg the pendingTransports array was getting disconnected by disconnectAllTransports, as each disconnection (in deactivateTransport) was mutating pendingTransports to remove itself. This would be ok if the removal used `delete`, but it (meaning Utils.arrDeleteValue) used slice, which renumbers as it goes, so array iteration gets messed up.
Configuration menu - View commit details
-
Copy full SHA for 68a106f - Browse repository at this point
Copy the full SHA 68a106fView commit details -
Rewrite reauth_authcallback test to check capability, not ttl
So it doesn't fail if the token takes too long to get, and/or take ages to give it enough buffer not to do the other thing
Configuration menu - View commit details
-
Copy full SHA for 7173dda - Browse repository at this point
Copy the full SHA 7173ddaView commit details -
Why? Because you can't abort requests. (You can remove the node, but that doesn't cancel the request). So recv requests stick around for 90s till realtime ends them. So in a test, the browsers max-connections-per-host limit fills up quickly, which messes up other comet transports too
Configuration menu - View commit details
-
Copy full SHA for ca9ab91 - Browse repository at this point
Copy the full SHA ca9ab91View commit details -
TEMP NOTFORMERGING use staging rather than sandbox for ci
While sandbox is unstable, to avoid spurious failures
Configuration menu - View commit details
-
Copy full SHA for 496d268 - Browse repository at this point
Copy the full SHA 496d268View commit details