Dmitriy Sobolev

These are the gaps which remain after the main review (#1257): 1. Separate declaration and definition (can be postponed until productization): See discussion: https://github.com/oneapi-src/oneDPL/pull/1257#discussion_r1407746163. 2. Use `zip_iterator`/`zip_view` instead of `__rng_pack`:...

These are the gaps which remain after the main review (https://github.com/oneapi-src/oneDPL/pull/1257): 1. Use global block load functions instead of gather when possible. See discussion: https://github.com/oneapi-src/oneDPL/pull/1257#discussion_r1414570090 and https://github.com/oneapi-src/oneDPL/pull/1257#discussion_r1422767127 2. Shrink types:...

Currently, oneDPL does not have such tests. It might be useful to verify ODR compliance.

This definition should be limited once the fix is available (it is not available as of 20240000 version): https://github.com/oneapi-src/oneDPL/blob/234b3ae0a5ad46ad057f1cc50b204e33a87f12ef/include/oneapi/dpl/pstl/onedpl_config.h#L193-L199

The logic of generation of CMake and pkg-config files can be greatly simplified after #1062. There are no `linux` or `windows` subfolders in the updated layout. Probably, it worth removing...

Develop an approach to limit a broken macro as soon as the fix is available

2

As an intermediate step, follow https://github.com/oneapi-src/oneDPL/pull/1182#discussion_r1328580730 advice in order not to skip a version with a fix.

Let's test if we allow the compiler to generate unique kernel names for a default kernel name case.

[oneDPL] Mismatch between specified default comparator and algorithm signature

1

Let's consider `lower_bound` as an example: ```c++ template OutputIt lower_bound(Policy&& policy, InputIt1 start, InputIt1 end, InputIt2 value_first, InputIt2 value_last, OutputIterator result, Comparator comp = std::less()); ... If no comparator is...

Do not expect a compiler to define SYCL_LANGUAGE_VERSION

3

SYCL specification allows SYCL to be implemented purely as a library: https://registry.khronos.org/SYCL/specs/sycl-2020/html/sycl-2020.html#_library_only_implementation. It means that inclusion of `sycl/sycl.hpp` is a portable way to get `SYCL_LANGUAGE_VERSION`, in contrast to expecting it...

**Summary:** The issue mentioned in the release notes has been fixed on the IGC side: > esimd::radix_sort and esimd::radix_sort_by_key kernel templates fail to compile when a program is built with...