[SYCL] Add size_type, to accessor and local_accessor classes.

[mmoadeli]

• Add size_type to accessor and local_accessor classes.

[steffenlarsen]

Could you please add a small in-tree LIT test making sure size_type is available for the various accessors and that it is the right type?

[mmoadeli]

Could you please add a small in-tree LIT test making sure size_type is available for the various accessors and that it is the right type?

Thanks @steffenlarsen , done.

[bader]

@mmoadeli, OpenCL CPU and ESIMD emulator results were invalid due to infrastructure issue. I'll restart the pre-commit to get relevant results as soon as possible. No actions from your side are required. Sorry for inconvenience.

[steffenlarsen]

LGTM!

[bader]

Shouldn't we use C++ type (i.e. std::size_t) instead of C type (i.e. size_t)?

[mmoadeli]

Shouldn't we use C++ type (i.e. std::size_t) instead of C type (i.e. size_t)?

Specification Interface for buffer command accessors defines it as size_t.

[bader]

Shouldn't we use C++ type (i.e. std::size_t) instead of C type (i.e. size_t)?

Specification Interface for buffer command accessors defines it as size_t.

Okay. Let me forward this question to @intel/dpcpp-specification-reviewers.

[gmlueck]

Shouldn't we use C++ type (i.e. std::size_t) instead of C type (i.e. size_t)?

Specification Interface for buffer command accessors defines it as size_t.

I agree that the SYCL spec should use the std namespace when referring to size_t, and I opened an issue for this:

https://gitlab.khronos.org/sycl/Specification/-/issues/648

It would be better to use std::size_t in the CTS also.

FWIW, all C++ compilers that I know of recognize size_t (without namespace) as an alias to std::size_t even though the C++ spec does not guarantee this.

[AlexeySachkov]

********************
Unresolved Tests (1):
  SYCL :: AsyncHandler/default_async_handler.cpp

********************
Timed Out Tests (4):

3 warning(s) in tests
  SYCL :: KernelAndProgram/spec_constants_after_link.cpp
  SYCL :: Regression/kernel_bundle_ignore_sycl_external.cpp
  SYCL :: Regression/same_unnamed_kernels.cpp
  SYCL :: Regression/static-buffer-dtor.cpp

Last 4 tests were disabled on HIP in intel/llvm-test-suite#1639, async_handler is reported in #8555