Testbench checking with a mock simulator?

One of the main differences between cocotb and traditional verification setups is that the cocotb testbench may have a runtime error that only shows up after simulating to this point, which may be hours or days later. I remember (but cannot find again) a recent discussion about cocotb vs. SV and this was the major disadvantage for some people.

I wonder if we could find some of these problems early. One thing that came to mind is having some kind of a mock simulator that “immediately” returns from a Timer yield, immediately triggers a trigger, … This would not cover all code paths, but may still be useful. Or is there a way to statically check the code for (some of) the potential problems?

[eric-wieser]

Note that to enable users to get the most out of the mypy linter, cocotb would have to opt in and provide the necessary annotations. Probably something that would be much easier after a python 3 transition.

[cmarqu]

Superseded by other PRs.