Details
- Reviewers
ldionne - Group Reviewers
Restricted Project - Commits
- rG8af259e8d90b: [libc++][PSTL] Implement std::{any, all, none}_of
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
libcxx/test/std/algorithms/alg.nonmodifying/alg.any_of/pstl.any_of.pass.cpp | ||
---|---|---|
29 | Like other algorithms, this should be tested on the different iterator categories. |
This is nice! I would suggest refactoring the serial backend so that we can enable this with -fexperimental-library sooner rather than later, maybe after stamping out a couple more algorithms just to confirm there are no unknowns.
libcxx/include/__algorithm/pstl_any_all_none_of.h | ||
---|---|---|
29 ↗ | (On Diff #517389) | Can you please make sure you have tests for all nodiscard extensions you're adding? |
35 ↗ | (On Diff #517389) | I suspect we might end up removing those, but until then, we should keep the code self-descriptive. |
libcxx/include/__pstl/internal/parallel_impl.h | ||
69 ↗ | (On Diff #517389) | Let's be explicit about the type when we pass it. We can remove it later if not needed. |
libcxx/include/__type_traits/is_execution_policy.h | ||
38–39 ↗ | (On Diff #517389) | Let's return auto const& instead, this is kinda confusing for no benefit. |
libcxx/test/std/algorithms/alg.nonmodifying/alg.all_of/pstl.all_of.pass.cpp | ||
36 | Let's test on a few sizes of sequences: 1-element, 2-elements, and more than e.g. 100 elements. Not to test the backend itself (which will require its own tests per our discussion), but to test the interaction between the "front-end" of the API and the backend. |
I would suggest splitting the PSTL all_of, any_of and none_of into a separate header to avoid impacting the dependencies of the basic any_of algorithm.