Ashley Anderson

Yeah it's a bit sad but it would be more consistent behavior, even if the discrepancy would be rare. Still I'll defer to your (and others') expertise and experience on...

> Well @aganders3 nobody has raised concerns about also changing sequences, so I guess let's move ahead with that. Do you want to add to this PR, or shall I...

Thanks again for the review. It was trickier than I thought to change the static to hold a `PyResult`, but I'm getting a better feel for result and error types...

> Thanks, this looks really solid. Just a couple of final things to decide on before we merge this in. I've commented on the sequence implementation, the equivalent ideas apply...

This seems related to the request here https://github.com/pyca/cryptography/pull/6695#discussion_r767289574 as well? @messense can correct me but it seems they want to compile another static lib with the same version of Python...

The [wurlitzer](https://github.com/minrk/wurlitzer) Python package may help in the shorter term. It can be loaded as an IPython extension.

I finally got coverage working locally and I'm going to chip away at some of these.

My experience is more in Python but most projects I've worked with exclude tests from coverage. I'm not sure if this is just convention or best practice but I'd agree...

I see why this is confusing but this happens because the CWD is changed to a tmp dir with _only the contents of the input wheel_ when entering this [`InWheelCtx`...

I can't seem to attach a wheel (or a zip of a wheel) to this post, so here is a link: http://files.aga3.xyz/antspyx-0.3.2-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl