GeezerVisonThe World Wide Web, restored to 800 pixels of glory. | ||||||||||||
| Home|About|Gaming cabinet|Archives|How it works|Starter sites | ||||||||||||
GeezerVison is showing github.com — GH-97696: Initial work on eager tasks by gvanrossum · Pull Request #98137 · python/cpython Conversation
|
|
|
||
| return task | ||
|
|
||
| def eager_task_factory(self, coro, *, name=None, context=None): |
There was an error while loading. Please reload this page.
There was a problem hiding this comment.
The reason will be displayed to describe this comment to others. Learn more.
Hm, this doesn't have to be a loop-level method? I imagined this to be a top-level function in the asyncio package.
One would install it like this:
async def main(): loop = asyncio.get_running_loop() loop.set_task_factory(asyncio.eager_task_factory) # ... the rest of the main() function ... asyncio.run(main())
Sorry, something went wrong.
There was an error while loading. Please reload this page.
There was a problem hiding this comment.
The reason will be displayed to describe this comment to others. Learn more.
I didn't try very hard to make that work yet... The problem is that the new factory has to call the old factory, but (at least in the Python version) the old factory defaults to None and then create_task() just constructs an instance of the Task class. But when there's a different loop I don't know how to do the equivalent thing -- can I just assume that get_task_factory() shouldn't return None if they have a different task class? (Is uvloop's factory ever None?)
Sorry, something went wrong.
There was an error while loading. Please reload this page.
There was a problem hiding this comment.
The reason will be displayed to describe this comment to others. Learn more.
The problem is that the new factory has to call the old factory
Why? I don't think this is a requirement. I remember we discussed it with the instagram people and the idea was to have two functions in asyncio:
# The custom eager factory API that instagram wants to use because # they have a bunch of instrumentation on top of asyncio.Task: def asyncio.create_eager_task_factory(custom_task_constructor): def factory(coro): try: result = coro.send(None) except StopIteration as si: f = Future() f.set_result(si.value) return f except Exception as ex: f = Future() f.set_exception(ex) return f else: return custom_task_constructor(coro, coro_result=result) return factory # The default eager factory API that most users will use: asyncio.eager_task_factory = create_eager_task_factory(asyncio.Task)
The implementation of loop.set_task_factory and loop.create_task methods would stay the same. We'd have to add the coro_result= keyword-only argument to asyncio.Task (both Python and C versions), but that I think is acceptable.
I just assume that get_task_factory() shouldn't return None if they have a different task class? (Is uvloop's factory ever None?)
I think None is just a marker for the "default factory is used" scenario. I don't think we were ever wanted to chain factories.
Overall I really want this to be implemented not on the event loop, the loop API surface is already very large.
Sorry, something went wrong.
There was an error while loading. Please reload this page.
There was a problem hiding this comment.
The reason will be displayed to describe this comment to others. Learn more.
Oh, that's a nice solution. If I get back to this before anyone else I'll use that.
Sorry, something went wrong.
There was an error while loading. Please reload this page.
…ng to get it working
…ng to get it working
…ng to get it working
There was an error while loading. Please reload this page.
Successfully merging this pull request may close these issues.
There was an error while loading. Please reload this page.