Method Definition Possible Issue JACC v0.1.0

[alexishuante]

Hello, I received this warning on precompilation when using JACC v0.1.0:
┌ JACC → JACCONEAPI
│ WARNING: Method definition parallel_for(oneAPI.oneAPIKernels.oneAPIBackend, Tuple{I, I}, F, Any...) where {I<:Integer, F<:Function} in module JACCONEAPI at /home/xa2/.julia/packages/JACC/zbR1L/ext/JACCONEAPI/JACCONEAPI.jl:28 overwritten at /home/xa2/.julia/packages/JACC/zbR1L/ext/JACCONEAPI/JACCONEAPI.jl:41.
│ ERROR: Method overwriting is not permitted during Module precompilation. Use __precompile__(false) to opt-out of precompilation.

It seems it is pointing to these specific lines of code in JACC.jl/ext/JACCONEAPI/JACCONEAPI.jl:
function JACC.parallel_for(
::oneAPIBackend, (M, N)::Tuple{I, I}, f::F, x...) where {
I <: Integer, F <: Function}
maxPossibleItems = 16
Mitems = min(M, maxPossibleItems)
Nitems = min(N, maxPossibleItems)
Mgroups = ceil(Int, M / Mitems)
Ngroups = ceil(Int, N / Nitems)
oneAPI.@sync @oneapi items=(Mitems, Nitems) groups=(Mgroups, Ngroups) _parallel_for_oneapi_MN(
(M, N),
f, x...)
end

function JACC.parallel_for(
::oneAPIBackend, (L, M, N)::Tuple{I, I}, f::F, x...) where {
I <: Integer, F <: Function}
maxPossibleItems = 16
Litems = min(M, maxPossibleItems)
Mitems = min(M, maxPossibleItems)
Nitems = 1
Lgroups = ceil(Int, L / Litems)
Mgroups = ceil(Int, M / Mitems)
Ngroups = ceil(Int, N / Nitems)
oneAPI.@sync @oneapi items=(Litems, Mitems, Nitems) groups=(
Lgroups, Mgroups, Ngroups) _parallel_for_oneapi_LMN((L, M, N),
f, x...)
end

The warning was gone once I changed (L, M, N)::Tuple{I, I} to (L, M, N)::Tuple{I, I, I} for the 3D parallel_for. I looked at ext/JACCAMDGPU.jl and ext/JACCCUDA.jl and it seems they have set as (L, M, N)::Tuple{I, I, I} for the 3D parallel_for. Not sure if this is the intended way to do it.

[PhilipFackler]

@alexishuante, our oneapi backend has been neglected to say the least. Today I submitted a PR that gets our tests working with oneapi. In the process, I believe I ran into this same error and fixed it. If you don't mind, could you try out the version in the PR #151 and let me know if this is resolved? Thank you.

[alexishuante]

Yes, I can confirm the issue has been resolved, thank you.