Graham Markall

> Maybe we need to set these attributions read-only if we cannot trace the changes. That's a good idea - I'll have a quick look into how difficult it would...

> A example of changing superclass attributions: Also, I can't really see any reference to inheritance in the docs (https://numba.readthedocs.io/en/stable/user/jitclass.html)- however there are some tests of inheritance in the test...

> I'll have a quick look into how difficult it would be to implement that. I haven't been able to figure out how to implement it in the time available...

In the end I couldn't stop myself from trying. The following seems to prevent updating the members of a jitclass without breaking the jitclass tests: ```diff diff --git a/numba/experimental/jitclass/base.py b/numba/experimental/jitclass/base.py...

Summary from the triage meeting - preventing the updates in this manner appears to be a reasonable thing to do.

Removed the stale label. @Mskxn Do you have any bandwidth to contribute a PR based on the above patch and the addition of an appropriate test?

Thanks for the report! I can reproduce this.

Many thanks for the report and looking into the workaround. I agree this is not a good situation, so will add this to the 0.60 milestone to aim for a...

> Didn't read CUDA/NVVM documentation right now, I bet NVVM also provides 0, 1, 2, 3? NVVM only supports 0 and 3... I think if the user sets 1 or...

This should be resolved by nvidia/numba-cuda#8 (which will be the default implementation of the target as per https://numba.discourse.group/t/rfc-moving-the-cuda-target-to-a-new-package-maintained-by-nvidia/2628/4). The fix has gone into the `develop` branch there for now, but...