<algorithm> Implement P2302R4 ranges::contains #2911
Conversation
SuperWig
changed the title
stl/inc/algorithm
Outdated
| requires indirect_binary_predicate < ranges::equal_to, projected<_It, _Pj>, | ||
| const _Ty* > // | ||
| _NODISCARD constexpr bool operator()(_It _First, _Se _Last, const _Ty& _Val, _Pj _Proj = {}) const { | ||
| return _RANGES find(_STD move(_First), _Last, _Val, _Pass_fn(_Proj)) != _Last; |
There was a problem hiding this comment.
No change requested: there are _Pass_fn calls at multiple layers here, which we usually avoid, but I checked and _Pass_fn will not wrap twice, so this seems reasonable for now.
|
Thanks! I validated and pushed minor fixes/cleanups, the most significant of which was guarding the product code for C++23. |
|
I'm speculatively mirroring this to the MSVC-internal repo (because I'm a greedy kitty who likes treats) - please notify me if any further changes are pushed. |
|
@StephanTLavavej pushed a minor change to the |
|
@SuperWig Updated the mirror, thanks for letting me know! 😻 |
|
@StephanTLavavej I think there are some issues that should be addressed |
| requires indirect_binary_predicate<ranges::equal_to, projected<iterator_t<_Rng>, _Pj>, const _Ty*> | ||
| _NODISCARD constexpr bool operator()(_Rng&& _Range, const _Ty& _Val, _Pj _Proj = {}) const { | ||
| // clang-format on | ||
| const auto _UResult = _RANGES _Find_unchecked(_Ubegin(_Range), _Uend(_Range), _Val, _Pass_fn(_Proj)); |
There was a problem hiding this comment.
I believe we should memoize _Uend(_Range) as calling it twice might be expensive
There was a problem hiding this comment.
We generally don't avoid repeated calls to begin and end, because they are generally O(1). The first call may be expensive, but the subsequent calls are not. (Caching things across big function calls consumes stack and pushes us closer to cold stack space, so it's not a no-brainer.)
| return true; | ||
| } | ||
|
|
||
| const auto _Match = _RANGES search(_Range1, _Range2, _Pass_fn(_Pred), _Pass_fn(_Proj1), _Pass_fn(_Proj2)); |
There was a problem hiding this comment.
Should we forward _Range1 and move _Range2
There was a problem hiding this comment.
lvalue ranges are strictly more capable than xvalue ranges, so forwarding is just noise for the purposes of algorithms.
|
Thanks, updating mirror! 🪞 |
|
Thanks for implementing this C++23 ranges feature - I can barely contain my excitement! 😹 🎉 😻 |
|
Glad to be part of bringing this super advanced technology that took decades to standardise. |
Co-authored-by: Stephan T. Lavavej <[email protected]> Co-authored-by: Casey Carter <[email protected]>
Fixes: #2921
I more or less just copied the tests for find and search. I'm guessing the feature test macro.