Stephan T. Lavavej

We're having a meeting on 2025-05-07 with the compiler and UCRT teams to figure out the plan here.

Not yet, the compiler team is still looking into this before deciding on a course of action. A major concern is how to address accuracy issues (i.e. should the compiler...

That's what I'm advocating for whenever the compiler team asks me.

It appears that this is, by definition, powered by compiler magic. :magic_wand:

Sure, I've moved this to Investigating for you. (As I've mentioned elsewhere, I do plan to open up the repo to C++26 PRs soon, although ideally we should land `feature/flat_map`...

Yes, I believe it is permissible to call `to_chars` via "as if". It's because there's no way for users to affect how `format` calls `to_chars` here.

I'll move this to Investigating to mark that you're interested. I plan to open up the repo to C++26 PRs in the near future, I just want to land `flat_meow`...

`exchange()` is an interesting case: https://github.com/microsoft/STL/blob/8f9431931b1e1f02d24d518452ae447a843c6121/stl/inc/utility#L568-L573 It has the side effect of performing an assignment, but if someone discards the return value, then it's just more expensive syntax for an...

Related: #1742 is a case where marking `lock_guard` as `[[nodiscard]]` warns for valid code like `m_val{ (std::lock_guard{my_mutex}, obj.m_val) }` in a *mem-initializer-list*. Although we remain very cautious about false positives,...

We talked about this at the weekly maintainer meeting and we believe that our implementation is following the Standard's vaguely mumbled guidance here. We're relabeling this issue as "LWG issue...