> So let's retarget this to 3.8 as it's too close to 3.7 feature freeze to get seriously considered/merged.

I *think* I have a better idea than this for 3.8, but it's definitely PEP-sized. The unawaited coroutine tracking proposal here was more conceived as a small, unobtrusive thing that could be added to 3.7 and marked provisional, to hold us over until 3.8. (Just this week I've spent time helping two people who were having trouble with forgetting 'await's, and this feature would have given them better errors -- though one was also affected by bpo-32703.)  OTOH the issue is complicated enough that it's entirely possibly my "better idea" has missed some showstopper detail, in which case maybe this is the best solution after all. We'll have plenty to talk about at PyCon this year :-).

Also, looking at the issue log here there's a ton of missing context from all the discussions that happened elsewhere this year and eventually led to this particular proposal; I'll try to make a list of links and post it here for future reference and archaeologists.