Hacker News new | ask | show | jobs
by red_admiral 703 days ago
eBPF is not Turing-complete, I suppose.
2 comments

It is not, programs that are accepted are proved to terminate. Large and more complex programs are accepted by BPF as of now, which might give the impression that it's now Turing complete, when it is definitely not the case.
In this talk we demo Conway's Game of Life implemented in eBPF: https://www.youtube.com/watch?v=tClsqnZMN6I
I should clarify that individual eBPF programs have to terminate, but more complex problems can be solved with multiple eBPF programs, and can be "scheduled" indefinitely using BPF timers