WIP Make kernel message handling async and in order #4697
Commits on Jun 12, 2018
-
First pass at making all kernel message handling asynchronous and pre…
…serving message order. Fixes jupyterlab#4188
-
First pass at handling the case for restarting a kernel and canceling…
… processing of old kernel messages in the async queue.
-
-
-
Commits on Jun 13, 2018
-
-
-
-
Rename checkCurrentMessage to assertCurrentMessage since it behaves m…
…ore like an assert.
-
Convert kernel utility functions to use async/await syntax.
This also fixes at least one bug, where the shutdownAll was resolving before each shutdown actually resolved.
-
-
-
-
Commits on Jun 14, 2018
-
-
-
-
Update the current kernel session id on *every* message.
The kernel session id is stored to make it easy to see if a previously scheduled message should still be processed, so it should be updated every time we get anything from the kernel. This fixes problems we were seeing where the messages updating the kernel session id were being ignored because they were being canceled.
-
Commits on Jun 15, 2018
-
-
-
Prevent console error when a future is canceled.
From @afshin, who made the fix in jupyterlab#4733.
Commits on Jun 16, 2018
-
-
-
Use phosphor’s defer rather than requestAnimationFrame
The test setup does not have requestAnimationFrame, so there is an error by assuming it.
-
-
-
-
It looks like we disabled all tests except the ikernel tests in 9b654f5 (see jupyterlab#4115, jupyterlab@9b654f5#diff-388fcef0ff8fa826feaa68effbcaaf02R30). We forgot to enable them again.
Commits on Jun 17, 2018
-
Adjust logging to use a single exported function.
This makes it easier to turn on and off logging while we fix the remaining tests.
Commits on Jun 18, 2018
Commits on Jun 19, 2018
-
-
Skip two tests (for now) that rely on kernel status messages in a par…
…ticular pattern.
-
-
Fix the serverSettings unit test.
We were getting an error about the baseUrl not being an absolute url (apparently a requirement of the node fetch library).
Commits on Jun 21, 2018
-
-
-
Remove the anyMessage ‘different client session’ test.
The anyMessage signal should be emitted for any message received, not just messages with a particular parent header.
-
-
-
-
-
-
-
-
-
-
-
Tweak TODO warnings on tests that sometimes time out.
Commits on Jun 22, 2018
-
Make a copy of the runningKernels list if there is a possibility of d…
…eleting one from the list.
-
Don’t preemptively cancel messages from a different kernel session.
This is so that code waiting on, for example, a status message will still resolve, especially in the case that the messages were buffered into another session. Any message handling that may require the current session (such as creating comms) should check to see if the message being handled is current.
-
Work around issues where tests were sporadically hanging.
It’s still a bit puzzling what is happening - it appears that the notebook is dropping messages from the kernel. More details at jupyter/notebook#3705.
-
-