Takayuki Matsuoka

@hltj thanks! You're right. I've fixed my comment to prevent further misunderstanding.

> Why are these incompatibilities allowed, but others aren't? As for `-Wno-long-long`, because we need this feature. But we also want to be friendly with C89/C90 as much as possible...

> Should this be removed? Since many of them can be compiled with `-std=gnu90`, it'd be nice to leave them as is. > We don't have a make c_standards_c17 test....

I made a new milestone `v1.9.5` and add some issues. https://github.com/lz4/lz4/issues?q=is%3Aissue+milestone%3Av1.9.5 If you notice/think any other issues which should be addressed as a part of `v1.9.5`, please tell us about...

chore: Since we're progressing to the next release, I've added #1372 and #1413 to [milestone:v1.9.5](https://github.com/lz4/lz4/issues?q=is%3Aissue+milestone%3Av1.9.5) to ease to make the next release note.

Cirrus CI failed to update the FreeBSD repo.

> Windows maybe ? Yes. I mainly think about (next release of) MSVC. Since the difference between `pthread.h` and `threads.h` is smaller than vs WIN32.

@leafi, thank you for pointing this out! You're right. Their return value compatibility must be checked as you suggested. Or all macro must return pthread compatible value.

> Hey, might it be worth including some kind of assert (static_assert? too new C std?) to make sure `thrd_success` from threads.h actually equals 0? dada21b addressed this issue.

I've tested new WIP `asyncio` branch a bit on Windows. Even on WSL2, it works great. - Target file : the almost incompressable file [ggml-alpaca-7b-q4.bin](https://github.com/ggerganov/llama.cpp) - Target environments - WSL2...