POSIX has become the standard for operating systems abstractions and interfaces over the decades. Two drivers for the design of the abstractions are the hardware constraints and the use cases of the time. Today, the speed balance between I/O and compute is shifting in favor of I/O, which is partly why coprocessors and special-purpose accelerators are becoming more mainstream. Therefore, we argue that the POSIX era is over, and future designs need to transcend POSIX and re-think the abstractions and interfaces at a higher level. We also argue that the operating system interface has to change to support these higher level abstractions.
POSIX is very entrenched. Whatever would replace it needs to be better enough to be worth the effort. That may never happen. However, if it does, I’m very curious about what the solution looks like.