goodboy

@njsmith so here's another fun one: If you enter a `breakpoint()` and then *continue*, ctrl-c handling entirely breaks from that point forward in `trio`. More fun is that it's entirely...

Oh wow wait ma bad @smurfix, just realized I almost always have `pdb++` installed. With the stdlib's debugger it's what you posted: ```python 3.9) >>> python trio_pdb.py > /home/goodboy/repos/tractor/trio_pdb.py(10)trace() ->...

Oh, and just so that I'm not talking lots of air, I **did** fix this in `tractor` by tweaking these settings, and in that case it's working in subprocesses as...

@mikeengland perfect, yes this is the default behavior I would expect. I think this adds fuel for this change I've got in mind.

As a follow up to the discussion around *cloned channels* versus *separate nurseries*, I had the use case (from [a test](https://github.com/tgoodlet/tractor/blob/master/tests/test_streaming.py#L84)) where the consumer task is the *lone* task *inside*...

@belm0 is this `FrontRunnerBroadcastValue` code published and/or refined elsewhere? I currently have a need for it (or something similar) in `tractor`.

@basak @belm0 thanks for both links; will report back when we finally get to adding a more formal api for this on IPC streams.

We have what appears to be a `tokio` style implementation of the *fastest consumer* style as a `broadcast_receiver()` wrapper around a `trio.abc.ReceiveChannel`. It's extremely nascent and in prototype testing atm;...

About to land goodboy/tractor#229 but more then happy to contribute all of the `tractor._broadcast.py` module to `trio` core if there's interest. Pretty happy with the performance and ease of use...

Also I'm doing some work to get it all running on python 3.8 so if you're cool with a PR for that I'd be glad to submit :)