Marlon James

VHPI and VPI iterators make use of the "caching" functionality when creating pseudo-regions. The pseudo-region object re-uses the handle of the parent region object that contains the generator statement. https://github.com/cocotb/cocotb/blob/924f35a3b7d39543118b7bfaed77dd4808e6612b/cocotb/share/lib/vhpi/VhpiCbHdl.cpp#L1152-L1165...

We may be able to catch `$error()` with the `cbError` callback.

A test like that could currently use (ugly) `expect_error=SimFailure` to pass.

This is the same problem from #2941. Both are Windows + conda. One is Riviera, the other is Questa.

Thanks for figuring that out! Is that `sh` you are running from a conda package?

I think it's worth considering the feature of using the `vpi_printf`/`vhpi_printf`/`mti_PrintMessage` functions for logging. See https://github.com/cocotb/cocotb/issues/2015#issuecomment-1027304668

Yes, I'll try to find time for this soon.

I started on this, but I ran into issues with a test for the callbacks. I had a test working with iverilog but when I ported it to the verilator...

There are several things to consider, and several ambiguities within the standard. There is absolute vs relative time as mentioned earlier ( steveicarus/iverilog#490 ). Also, because there are several timed...

These should be one-time callbacks. When registering for a callback, the VPI application can provide a time delay before the callback is called, allowing scheduling callbacks for the future. Verilator...