Kaleb Barrett

The VPI error messages are #4841. It complains when certain VPI functions are used during the compilation step. We might be able to prevent those (harmless) errors from being reported...

@facebreeze VCS 2018 is *very* old. There's little doubt to me it won't work. We can't always support old versions of tools due to the litany of issues they tend...

The error messages are https://github.com/cocotb/cocotb/issues/4866#issuecomment-3168888964. I'll try removing cocotb from the compilation phase rn. EDIT: It's not happy about it. It inexplicably crashes. I'll have to debug.

Not sure right now. Apparently loading the VPI during compilation, even though it does nothing, has some effect. VCS is doing something very weird I haven't figured out yet.

I have a funny feeling that this is exposing a bug in Clock.

This is the weirdest thing. The tests seem to fail randomly. I can't replicate locally, but that could be because Python 3.12 works for some reason? I don't know what...

I think the issue was resolved by not adding the Task a second time calling `start_soon` on the result of `clock.start`.

Worked around the issue by adding a small Sphinx extensions to filter out undocumented special methods. https://github.com/ktbarrett/coconext/pull/78

Are you planning on adding support to the Makefiles or the Python runner for this feature? Odd numbers on Verilator are unreleased versions, so there isn't any way to test...

The CI was updated to include Verilator v5.042, you should be able to test it now. As for Makefile and Runner support, we haven't done anything with adding support for...