[SYCL] fix a typo in the comment

[jinz2014]

No description provided.

[steffenlarsen]

LGTM! NFC so failures are unrelated.

[jinz2014]

Is a SYCL buffer of size 0 fine ?

  // Accessor with buffer size 0.
  {
    try {
      int data[10] = {0};
      {
        sycl::buffer<int, 1> b{&data[0], 10};
        sycl::buffer<int, 1> b1{0};

        sycl::queue queue;
        queue.submit([&](sycl::handler &cgh) {
          sycl::accessor<int, 1, sycl::access::mode::read_write,
                         sycl::target::device>
              B(b, cgh);
          auto B1 = b1.template get_access<sycl::access::mode::read_write>(cgh);

          cgh.single_task<class acc_with_zero_sized_buffer>(
              [=]() { B[0] = 1; });
        });
      }
      assert(!"invalid device accessor buffer size exception wasn't caught");
    } catch (const sycl::invalid_object_error &e) {
      assert(e.get_cl_code() == CL_INVALID_VALUE);
    } catch (sycl::exception e) {
      std::cout << "SYCL exception caught: " << e.what();
      return 1;
    }
  }

[steffenlarsen]

@jinz2014 - Is this related to this patch? Either way, currently the spec does not disallow zero-sized buffers and does not block the construction of accessors from zero-sized buffers. This is an open issue on the spec however (KhronosGroup/SYCL-Docs#408) as this seems to be differing from the behavior of placeholder accessors on empty buffers, as it will throw when require is called on it.

[jinz2014]

No. I took a look at the unrelated failure. Thank you for explaining the issue.