extmod/uasyncio: Delay calling Loop.call_exception_handler by 1 loop, and add Task.done() method #6664
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
dpgeorge
force-pushed
the
extmod-uasyncio-delayed-detached-tasks
branch
from
9d35ee6 to
e405936
Compare
When a tasks raises an exception which is uncaught, and no other task await's on that task, then an error message is printed (or a user function called) via a call to Loop.call_exception_handler. In CPython this call is made when the Task object is freed (eg via reference counting) because it's at that point that it is known that the exception that was raised will never be handled. MicroPython does not have reference counting and the current behaviour is to deal with uncaught exceptions as early as possible, ie as soon as they terminate the task. But this can be undesirable because in certain cases a task can start and raise an exception immediately (before any await is executed in that task's coro) and before any other task gets a chance to await on it to catch the exception. This commit changes the behaviour so that tasks which end due to an uncaught exception are scheduled one more time for execution, and if they are not await'ed on by the next scheduling loop, then the exception handler is called (eg the exception is printed out). Signed-off-by: Damien George <[email protected]>
This is added because task.coro==None is no longer the way to detect if a task is finished. Providing a (CPython compatible) function for this allows the implementation to be abstracted away. Signed-off-by: Damien George <[email protected]>
dpgeorge
force-pushed
the
extmod-uasyncio-delayed-detached-tasks
branch
from
e405936 to
309dfe3
Compare
tannewt
added a commit
to tannewt/circuitpython
that referenced
this pull request
Jul 28, 2022
Add board.LED (alias of board.LED_R part of RGB)
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
When a tasks raises an exception which is uncaught, and no other task await's on that task, then an error message is printed (or a user function called) via a call to Loop.call_exception_handler. In CPython this call is made when the Task object is freed (eg via reference counting) because it's at that point that it is known that the exception that was raised will never be handled.
MicroPython does not have reference counting and the current behaviour is to deal with uncaught exceptions as early as possible, ie as soon as they terminate the task. But this can be undesirable because in certain cases a task can start and raise an exception immediately (before any await is executed in that task's coro) and before any other task gets a chance to await on it to catch the exception.
This commit changes the behaviour so that tasks which end due to an uncaught exception are scheduled one more time for execution, and if they are not await'ed on by the next scheduling loop, then the exception handler is called (eg the exception is printed out).
As part of this,
task.coro==Noneis no longer the way to detect if a task is finished, so theTask.done()method is added.IMO this fix is useful in its own right, but it's also a precursor to fix
wait_for()cancellation in a clean way (see also #5931).