Consider switching to building with zlib-ng about turbovnc HOT 3 OPEN

Switching our internal zlib implementation requires a lot of testing, so it isn't a trivial task. I've already done that once within the past six months (d016d0e, which updated our internal zlib implementation to the latest Intel zlib code), so I'm not in the mood to do so again right now. However, I will leave this open for consideration in TurboVNC 3.2. It would be nice to have SIMD acceleration for Arm CPUs, which zlib-ng appears to provide. However, I would need to thoroughly benchmark it to ensure that it provides significant enough speedup to justify itself on Arm CPUs and that it provides similar speedup to Intel zlib on x86 CPUs. (That includes SSE2-only CPUs, which TurboVNC still supports.)

from turbovnc.

I was already aware of it and have tested it in the past (#95), but that was some years ago. I'm sure it has improved since then.

from turbovnc.

I saw the commit, and I figured I'd highlight the modern project. The other idea of course is to stop directly integrating a versioned copy in the source and take advantage of perhaps submodules and their cmake integration:

https://github.com/zlib-ng/zlib-ng/wiki/Common-build-problems#using-zlib-ng-with-third-party-projects-in-cmake

from turbovnc.