Michał Chojnacki

+1 on @briansmith's experience. > (...) Ultimately the most important benchmark involves taking non-owned bytes and encrypting them (...) As it is currently, rustls **_owns_ the data _anyways_** due to...

> > As it is currently, rustls **_owns_ the data _anyways_** due to at least 2 (IIRC) unnecessary allocating copies of the appdata passed in > > Could you point...

@JoshuaBatty Would including... ``` const AUDCLNT_STREAMFLAGS_AUTOCONVERTPCM: DWORD = 0x80000000; const AUDCLNT_STREAMFLAGS_SRC_DEFAULT_QUALITY: DWORD = 0x08000000; const AUDCLNT_STREAMFLAGS_PREVENT_LOOPBACK_CAPTURE: DWORD = 0x01000000; ``` ... in `um/audiosessiontypes` be an option, while at it? The...

For the record, the issue lies in `cpal`, which does not set a high/real-time priority on its playback thread on Windows. (Was part of the reason why I decided to...

Aloha, I'm not sure I follow - is there some issue with the examples in the docs or in the README?

@lyrixx > Can you share some bench? It's a O(n) change to O(1), with an N of 55k+, where the initialization cost is paid only once, as opposed to the...

> The only consideration being `sendBeacon` ensures to request is made, regardless of navigation away from the page that initiates it, whereas fetch does not. I believe this is a...

> Yes, true, but mileage may vary between browser vendors! Not sure I follow. https://caniuse.com/?search=keepalive shows ~97% compat for a measure intended as a _fallback_ to sendBeacon with its 99%...