-
Notifications
You must be signed in to change notification settings - Fork 10
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fix find_spec for old finders that don't implement it. #86
Conversation
Marking as draft until we can figure out our merging/release strategy. Since numba-cuda has only released patch versions up until now, we will either need to add another release segment just for this case, or make a post release. RAPIDS 24.12 is pinned to <0.0.18 due to incompatibilities that introduces, see rapidsai/cudf#17359. |
Copying what I wrote offline... I think we should release this as |
Ah, crumbs. I rebased this to a branch starting on v0.0.17 and Github's now pulling in all the history from the main branch. |
My plan here was:
|
Rebased accordingly. |
The code in https://github.com/NVIDIA/numba-cuda/blob/v0.0.17/site-packages/_numba_cuda_redirector.py#L70 is incompatible with legacy finders (pre-Python 3.4) that do not implement the
find_spec
method. This makes numba-cuda incompatible with Google Colab (see googlecolab/colabtools#5000). For the purpose of RAPIDS 24.12, which is an early adopter of numba-cuda, we need this fix to go into a patch release of 0.0.17 because RAPIDS 24.12 is incompatible with numba-cuda>=0.0.18.