A. Jiang

`<xutility>`: Evil member `element_type` can break non-ranges algorithms since C++20

4

# Describe the bug Currently, MSVC STL defines `_Iter_value_t` in term of `iter_value_t` since C++20 and uses it in non-ranges algorithms. https://github.com/microsoft/STL/blob/8b081e26ba016970ce3338cb483ff10f2ade30a5/stl/inc/xutility#L1167-L1168 https://github.com/microsoft/STL/blob/8b081e26ba016970ce3338cb483ff10f2ade30a5/stl/inc/xutility#L1177-L1178 Per [[readable.traits]](https://eel.is/c++draft/readable.traits), if a type `I` has...

The new header `` provides `char_traits` and `basic_string_view` and related utilities. `` shouldn't provide other utilities in ``. Fixes #4625. `` seemingly still need to include `` in C++23 due...

Is it by design that `<string_view>` provides most contents of `<string>`?

1

MSVC STL's `` currently directly includes ``, which provides most contents of ``. I think it's OK for `` to include `` (since C++17). But when one only includes ``,...

`<utility>`: Side effects in self-swaps of `pair` are skipped

1

# Describe the bug Currently, self-swaps of `pair` are no-op: https://github.com/microsoft/STL/blob/9aca22477df4eed3222b4974746ee79129eb44e7/stl/inc/utility#L441-L460 However, the _Cpp17Swappable_ requirements ([[swappable.requirements]](https://eel.is/c++draft/swappable.requirements)) doesn't seem to exclude self-swaps or side effects in them, and no special case...

2.3 中的 `typeof` 不是 C++ 中的标准特性

1

## Actual Description - File Path: `book/en-us/02-usability.md` - Original paragraph: 2.3 ``` The `decltype` keyword is used to solve the defect that the auto keyword can only type the variable....

Improve no_destructor.h

5

1. Removed dependency on `std::aligned_storage` (which will be deprecated in C++23; see also https://github.com/cplusplus/papers/issues/197). 2. Removed redundant `static_assert`s. 3. Guaranteed the placement new-expression calls global `operator new`.

This may be still fine if we only accept stateless allocators. E.g. `std::allocator` is fine, but `std::pmr::polymorphic_allocator` is not. However, if so, we should require `allocator_traits::is_always_equal::value` to be `true` somewhere.

[[contents]/2](https://eel.is/c++draft/contents#2) already allows implementations to put the components into an inline namespace in `std`. So, it seems overspecified to require using an inline namepace of a specified name `json_v1`.

Not sure whether this is intended. And this isn't consistent with the implementation in `customization.h`. https://github.com/16bit-ykiko/magic-cpp/blob/8d352d71ffb87e7a6754ca0758205eee3a52040a/README.md#L42-L48 https://github.com/16bit-ykiko/magic-cpp/blob/8d352d71ffb87e7a6754ca0758205eee3a52040a/include/magic/customization.h#L11-L15 --- Also, the use of such a pointer is possibly (but not always)...

STL: `expected<any, T>` and its friends can break container iterator comparison

12

# Describe the bug The bug reported in [GCC Bugzilla #115939](https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115939) is currently also present in MSVC STL. Iterator comparison is currently broken for all but set containers. The reason...