CFLAGS etc.

[cloudflare]

CFLAGS etc.

[jdesgats]

You are right, we should not hard-code CPU-specific compile flags (being amd64 or ARM). I guess that a proper configure script or cmake file could detect them automatically, but we don't have that so far so your PR is the best way forward.

Sorry again, I ogt busy on many other things and I needed to check where this module is used in our internal repos in order to not have any regressions.

[vcunat]

There are also other weird things I see, e.g. passing CFLAGS to c++ compiler, etc. so I'd probably look at these as well.

[vcunat]

OK, I created some implementation instead of just talking: #13

[pspacek]

@vavrusa Could you please poke someone to get fixes merged? Thanks!

[vcunat]

Or just get some "authoritative" opinion; it's not a big deal for us to maintain a fork, but it would seem better if we can agree on one common version.

[jdesgats]

Hey, sorry for the late answer. It turns out we already fixed this issue internally, I pushed the corresponding patches in #14

[vcunat]

Well, this is a misunderstanding between us – my concern was that by default the code should work on any CPU of the given architecture. Old example: x86_64 does not guarantee sse4.1; new example: aarch64 does not guarantee the CRC instructions. I actually suspect that our device which already runs this code doesn't have CRC, but I've been primarily concerned about compatibility from the point of view of generic packaging for unix-like systems.

[jdesgats]

Solved by merging #13.