> This is exciting and I'm happy that you're addressing this problem. Thank you!
> Some of our use cases can't be implemented using this PEP; notably, we use a > timing context that times how long an asynchronous function takes by > repeatedly pausing and resuming the timer. Measuring performance of coroutines is a bit different kind of problem. With PEP 550 you will be able to decouple context management from collecting performance data. That would allow you to subclass asyncio.Task (let's call it InstrumentedTask) and implement all extra tracing functionality on it (by overriding its _send method for example). Then you could set a custom task factory that would use InstrumentedTask only for a fraction of requests. That would make it possible to collect performance metrics even in production (my 2c). Yury _______________________________________________ Python-ideas mailing list Python-ideas@python.org https://mail.python.org/mailman/listinfo/python-ideas Code of Conduct: http://python.org/psf/codeofconduct/