Dan Hoeflinger

@SergeyKopienko, I'd like to get your take on this. If there already exists something like this, or if you have ideas on how this would best augment our current testing.

oneDPL has made significant progress in adding SYCL device copyable specializations for oneDPL's iterator types and wrapper functions. There was a question about whether this removes the need for a...

> These additions LGTM. Do we need a documentation update as well listing the newly supported operators in `sycl_iterator`? Yes, we can update the guide with this on the supported...

@mmichel11 The operator support of the "unspecified type" is actually defined by the specification, so if we are updating the guide, we also need to look at updating the specification....

@adamfidel Thanks for taking a look. > Since in practice the `sycl_iterator` class provided `operator

To summarize an offline conversation about this PR: The oneDPL spec is quite sparse in its language defining custom iterators like zip_iterator, and about what requirements exist on the iterators...

With 1) The changes to `transform_iterator` in #1445, where we decided to keep copy assignment even when functors are not copy assignable (by skipping the assignment of the functor), 2)...

> The description says: > > > ... based on oneDPL device copyable tuple > > According to SYCL 2020, `std::tuple` is also device copyable (if the wrapped types are...

Perhaps @adamfidel is the better person to tag here. I'm happy to make some changes here to improve this, but I didn't write this code.

Do we want to name this something other than `*.pass` so it doesn't get included in our full test glob? The internal CI specifically builds and runs this test separately,...