Taylor Nie

Seems to be fixed in a new brave release (1.42.88)

I see that numba reimplements numpy functions in Python, then use the jit compiler to compile that. https://github.com/numba/numba/blob/main/numba/np/arraymath.py#L4079 I'm not super familiar with numba internals, but what's the status on...

@esc understood. What's Numba's plan forward for performance regressions like these, where numpy/scipy can use well-tuned native vectorized code? Are we just going to wait on improvements to the jit...

I found a better way to do this. Github Action actually lets you run jobs **inside** docker containers, rather simply using the entrypoint's of a given container (which is currently...

I don't like copying the docs of `os.rename` verbatim here. Should we just instead link to the docs of `os.rename` and say that's what `Path.rename` uses internally?

@tupui done https://mail.python.org/archives/list/[email protected]/thread/65RQPPV6MZLLLTJX4DB34BA5JWIR56RX/

@peterbell10 @mreineck Just wanted to loop you guys into this issue. Since `signal.oaconvolve` is implemented in pure python and is extremely slow (thereby defeating the purpose of the overlap-add method...

> I'm pretty sure you are seeing the effect of the larger calling overhead of both pocketffts in comparison to FFTW. This is unavoidable in designs where the user does...

Could you comment on the threading issue? I don't see any speed up with increasing `nthread` passed to `r2c` or `c2r` 

This seems like a regression introduced from the migration from coffee script. The [chrome version of vimium (v1.66) was last updated on 2 March 2020](https://chrome.google.com/webstore/detail/vimium/dbepggeogbaibhgnhhndojpepiihcmeb), while the [firefox version of...