-
Notifications
You must be signed in to change notification settings - Fork 232
Test with latest CCCL #2106
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
base: main
Are you sure you want to change the base?
Test with latest CCCL #2106
Conversation
|
Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually. Contributors can view more details about this message here. |
|
/ok to test |
|
/ok to test |
|
@miscco It seems like some issues are resolved but there are still build problems. I'm not sure if this is RMM or CCCL changes at fault. I will try to investigate further when I get a chance (dealing with other build breakage right now). |
050919f to
689d14c
Compare
|
/ok to test |
|
/ok to test |
2 similar comments
|
/ok to test |
|
/ok to test |
|
/ok to test |
Inside a class template like cccl_resource_ref<ResourceType>, the bare name cccl_resource_ref refers to the injected class name (the current instantiation), not the class template. The is_specialization_of_v trait requires a class template as its second argument, causing a type/value mismatch error. Use fully qualified names (::rmm::detail::cccl_resource_ref and ::rmm::detail::cccl_async_resource_ref) to bypass the injected class name and correctly refer to the class template.
Replace std::optional with cuda::std::optional from libcu++ in cccl_resource_ref and cccl_async_resource_ref. The standard library std::optional lacks CUDA annotations and cannot be used in __device__ or __host__ __device__ contexts, causing warnings when thrust_allocator (which has __host__ __device__ constructors) copies its resource ref member.
|
/ok to test |
This is a test PR to demonstrate an issue in building with the CCCL
mainbranch.