jaskij
jaskij
Sure thing. I wasn't sure if that quickly put together operator was good enough, but if it is, I'll go ahead and submit a PR.
cc @eivindbergem as this may be related to #58 - both revolve around dependency resoluton
@BKDaugherty thanks! That does look interesting. Sadly, I'm already using the version from #70 because of git dependencies.
> Would you expect integrals to be extracted at byte offsets only, or arbitrary offsets into the bitset? The use case it came up in was Modbus coil/discrete encoding/decoding, which...
> Byte boundaries could result in more optimal code. Since checking for byte boundaries is very cheap, maybe it would be worth the effort of writing both versions, and using...
Looking at the API, would you consider defaulting `length` to `sizeof(T) * CHAR_BIT`? It seems like it would be a good default, at least for the use case I've mentioned....
The API looks good to me then. I thought a little about extraction to, say, `std::output_iterator` or something, but in the end it unnecessarily expands the API surface for no...
Great! Thank you!
Mostly it's about code size. Depending on my own build configuration, I'm currently using over 90% of the allowed space. Better Enums without `BETTER_ENUMS_CONSTEXPR_TO_STRING` would push me over 100%, causing...
To be specific, this happens when trying to compile `pypocketfft.cxx` in `scipy/fft/_pocketfft`. It seems that the issue here is more with pybind11 than with NumPy or SciPy