09:37 <zyp[m]> Catherine: we need to talk about what problem `run_async()` is supposed to solve

09:38 <zyp[m]> I've been thinking about it, and the way I figure it'd work would be more or less functionally equivalent to calling asyncio.run() inside a testbench to await an asyncio awaitable, especially if you use nest-asyncio to make it reentrant

09:43 <zyp[m]> I figure if a testbench is waiting for data from some external source, e.g. await sock.recv() or whatever, it would in any case be blocking the whole simulation from advancing

09:45 <zyp[m]> and yeah, being able to pass an external awaitable from a testbench up through run_async() is a cleaner solution that running an io loop reentrantly, but is that the entire value proposition?

12:38 <whitequark[cis]> zyp: I don't think nest-asyncio is a legitimate proposal

12:38 <whitequark[cis]> it's some weird monkey patch. as far as I'm concerned that cannot be sued

12:38 <whitequark[cis]> s/sued/used/

12:38 <whitequark[cis]> so the problem it would be solving is "not requiring nest-asyncio"

12:39 <zyp[m]> I agree, just wondering if that's all there's to it, or if you had more in mind