Valentin Churavy

I am not a fan, but I understand where you are coming from. The semantics of environment variables has some long traditions. As an example ``` PATH="/a/b/c" PATH="$PATH:/a/b/c" PATH="/a/b/c:$PATH" ```...

No nothing immediate. I would run it through the [compute-sanitizer](https://cuda.juliagpu.org/stable/development/debugging/#compute-sanitizer) > related to how Enzyme.jl handles memory in CUDA kernels. No after a segmentation fault the device must be reset....

The primal code doesn't throw, so it might be something that is emitted as a fallback. Now I have a small enough example that I can stare at Monday

@wsmoses with Cthulhu: ``` ; │┌ @ /home/vchuravy/.julia/packages/KernelAbstractions/X5fk1/src/macros.jl:61 within `_mul!` ; ││┌ @ /home/vchuravy/.julia/packages/KernelAbstractions/X5fk1/src/KernelAbstractions.jl:779 within `construct` %12 = call ptr @jl_get_builtin_fptr(ptr nonnull @"+Core.#_compute_sparams#20514.jit") ```` ``` builtins.c 2181:JL_CALLABLE(jl_f__compute_sparams) 2183: JL_NARGSV(_compute_sparams, 1);...

@wsmoses adding `jl_f__compute_sparams` to the list is not effective... I assume this is because the pointer to name thing failed? Which is why the error message is not helpful?

None of the other `jl_f_` functions are listed there? And it doesn't help.

I mean this is a regression from 1.10 :)

Because you can't throw a method error using only the type of the function?

``` diff --git a/test/precompile.jl b/test/precompile.jl index e7a9a852..3c6e0830 100644 --- a/test/precompile.jl +++ b/test/precompile.jl @@ -42,8 +42,9 @@ precompile_test_harness("Inference caching") do load_path @setup_workload begin @compile_workload begin - autodiff(Reverse, mul, Active, Active(1.0), Active(2.0))...

> Also relatedly, can we use the inlineabi stuff on a temporary function to force Enzyme precompilation to compile all the interal utilities used by Enzyme [as otherwise presently the...