Philippe91

I have tried Windows/MSVC vs Apple/Clang on machines with similar processors, and Clang wins always (sometimes a lot). Worth noting: ska hashmaps beat robin_hood almost always. I guess (?) thanks...

> While most can definitely be canceled, changing the signature of all the wrappers would be an API incompatible change. I still need to figure out how to handle this.......

Did you consider running clang-tidy first, without fixes, and then to use the LLVM tool "clang-apply-replacements.exe" at the end, in a single thread? https://devblogs.microsoft.com/cppblog/exploring-clang-tooling-using-build-tools-with-clang-tidy/#lightbox-custom

@galchinsky On which CPU are you running? For example, Apple Silicon CPU behaves a bit differently than Intel, as I could experiment (SPSCQueue was not involved).

The stackoverflow link you mention is about support for aligned-new support, which is not the same thing as padding. Compilers automatically insert padding at the end of a struct, to...

I have added this compile-time assert to connect() to be on the safe side and keep best performances: static_assert(! std::is_lvalue_reference::value, "Use std::move() or construct{} the functor at call's location");

> I submitted a PR to fix this, see PR "Bug fix for #35". This solves the problem, apparently. This being said, the logic behind the cached indexes system introduced...

I think this is what David Stone proposes ("append_from_capacity") in his [CppCon talk](https://youtu.be/MfFzr9qqPDw)

On windows, use archive_read_disk_open_w and not archive_read_disk_open

The seed value is **only** used if the soxr output is 16 samples. In that case, some dithering process is applied. This means some noise is added to the signal...