Alexey Bader

> `jn/jnf` is probably an impl bug. @bader, thoughts? I haven't seen requests to support this functions before, so I'm not sure if we can classify this as an implementation...

> Besides my gcc version is gcc-9.5.0（build from source） I think some part of libstdc++ is not properly configured/built.

@Fznamznon, can we consider passing the whole SYCL kernel (lambda object) via single kernel parameter? This should allow reduce overhead on setting multiple kernel parameters.

> Although I don't see it as a thing that is not possible to do (assuming that we are targeting to pass the whole SYCL kernel object only when it...

Maybe we can add a compiler option to avoid the wrapping i.e. if user sets the flag it promises that all pointers captured inside device code are USM pointers.

> > Maybe we can add a compiler option to avoid the wrapping i.e. if user sets the flag it promises that all pointers captured inside device code are USM...

It's much easier to do in CFE where SPIR kernel function wrapper is created. I think maintaining address space adjustment in LLVM IR -> LLVM IR transformation or in SPIR-V...

Who will be responsible for `KernelFunctor` creation? The original test case is too simple to demonstrate all the questions we might have trying to make it work. If we add...

@intel/dpcpp-clang-driver-reviewers, I suppose clang driver adds additional include paths or system paths, where SYCL headers is found w/o additional directory prefix. Could you take a look, please?

I see following: "-internal-isystem" "bin/../include/sycl" "-internal-isystem" "bin/../include" I think we added "bin/../include/sycl" to enable `#include `, but it has a side effect, which described in this issue. To fix we...