asyncio
asyncio copied to clipboard
python
Tulip thread/process name analogue
This is a feature request, not a bug.
Thread names are a common logging aide in Python and other mainstream
languages. Since a particular Tulip event loop is inherently single threaded,
yet can still multiplex complex independent activities I propose a mechanism to
assign names to Tulip's thread analogues - Tasks/coroutines. This information
should then be accessible in some way by logger instances.
This should maybe be designed in a way that is useful to other async event loop
implementations, such as Twisted and Tornado?
Guido has expressed interest in the idea, but maybe not in the first release.
Original issue reported on code.google.com by [email protected] on 6 Oct 2013 at 1:09
Note that the name of the function used to start a Task is in its repr(). Maybe
the main missing thing is something that gets the current Task? Although I'm
not sure what to do if what's running didn't come from a task but is a handler
scheduled by e.g. call_soon().
Original comment by [email protected] on 8 Oct 2013 at 11:31
Original comment by [email protected] on 16 Oct 2013 at 4:42
- Added labels: Priority-Low
- Removed labels: Priority-Medium
We now have a class method Task.current_task() which you can use for this.
Original comment by [email protected] on 5 Dec 2013 at 11:36
Task.current_task() is analogous to threading.current_thread. The next step is
the analogue of threading.Thread.name.
The repr() might be the start of the solution, but we still need a counter to
disambiguate further (just like threading.Threads do). Maybe set the name at
creation time to "repr()-N"? The counter should probably be per-process instead
of per-loop too.
Original comment by [email protected] on 23 Feb 2014 at 1:14