Join GitHub today
GitHub is home to over 40 million developers working together to host and review code, manage projects, and build software together.
Sign upfix websocket broadcasts in Tornado 5.x #1796
Merged
Conversation
In Tornado 4.x and earlier, IOLoop.current() would return the global IOLoop if there was no IOLoop already running in the calling thread. This was the case when calling our websocket broadcast method from the HttpFrontend thread and so callbacks were correctly scheduled on the HttpServer thread's IOLoop. However, in Tornado 5.0+, the idea of a global IOLoop is gone and calling IOLoop.current() will simply create a new IOLoop if there isn't one already running in the calling thread. This incorrectly resulted in callbacks being scheduled on that new IOLoop which is never actually started and so the broadcasts were never sent. This is related to #1716.
kingosticks
force-pushed the
kingosticks:fix/tornado-5-support
branch
from
62b065a
to
de75e3f
kingosticks
force-pushed the
kingosticks:fix/tornado-5-support
branch
from
de75e3f
to
5af8fc7
This comment has been minimized.
This comment has been minimized.
|
Wow, the fact I managed to mess up that fixup and the tests still passed is really still a bit of an issue... |
This comment has been minimized.
This comment has been minimized.
|
Not entirely with you. Is it anything that should be fixed before merging? |
This comment has been minimized.
This comment has been minimized.
|
Nah, false alarm, it's good to go. |
Closed
jodal
added a commit
that referenced
this pull request
Sep 30, 2019
In Tornado 4.x and earlier, IOLoop.current() would return the global IOLoop if there was no IOLoop already running in the calling thread. This was the case when calling our websocket broadcast method from the HttpFrontend thread and so callbacks were correctly scheduled on the HttpServer thread's IOLoop. However, in Tornado 5.0+, the idea of a global IOLoop is gone and calling IOLoop.current() will simply create a new IOLoop if there isn't one already running in the calling thread. This incorrectly resulted in callbacks being scheduled on that new IOLoop which is never actually started and so the broadcasts were never sent. This is related to #1716. (cherry picked from commit 59a3935)
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
kingosticks commentedSep 8, 2019
In Tornado 4.x and earlier,
IOLoop.current()would return the global IOLoopif there was no IOLoop already running in the calling thread. This was the
case when calling our websocket handler's
broadcastmethod from theHttpFrontendthreadand so callbacks were correctly scheduled on the
HttpServerthread's IOLoop.However, in Tornado 5.0+, the idea of a global IOLoop is gone and calling
IOLoop.current()will simply create a new IOLoop if there isn't one alreadyrunning in the calling thread.
This incorrectly resulted in callbacks being scheduled on that new IOLoop
which is never actually started and so the broadcasts were never sent.
To fix this we explicitly provide
broadcastwith the correct IOLoop to use (HttpServer's).Also,
tornado.websocket.websocket_connectremoved the io_loop parameter, which we don't seem to need. I gave up trying to get the tests to fail under Tornado 5.x.This is related to #1716.