Shouldn't that regression already be handled by D155358?

Was caused by 3f3620e5c9ee0f7b64afc39e5a26c6f4cc5e7b37, thank you for fixing it up!

Thanks, this probably makes more sense than what I did before

Address comments

It appears other operators aren't affected, after D153641 I would have not been surprised if && had become unary :)

Add annotator test cases involving a lambda

@HazardyKnusperkeks I'm not sure why it didn't recurse already, given that you even documented that it doesn't, but I chose to trust in the Beyoncé rule.

Apply suggestions from review

This patch also fixes https://github.com/llvm/llvm-project/issues/52911 (which is probably a duplicate anyway)

Add annotator tests

Start iteration from second token

I see nothing wrong with this patch alone as it currently stands, since it's a quite simple change to the LineJoiner, and I see it as one of the final stopgaps in clang-format's support for requires clauses, and I think it definitely beats the current buggy behavior where behaviour is conflated with BraceWrapping.AfterFunction.
But I'd like other reviewers to maybe have another look?

(looks like you linked the same issue twice in the summary)

update commit message

Use MightBeFunctionDecl instead

This is great!

didn't go via the normal clang-format reviewers

Could this possibly be an issue for more esoteric underscore-less UDLs, like i?
Does the code need to search for a suffix, could it not detect the absence of a digit? Sorry if the questions are silly, I haven't really looked at this part of the formatter in depth

So, it took me a while but I finally found where the logic is that makes the lambda braces stay on one line, but, now I'm not so sure if I should change it:

Reduce column limit for macro test

In D145642#4213256, @usaxena95 wrote:

Another point:
While testing this patch, the following still fails to recognise. Might be something special with true.

auto y = [&]<typename Callable>
    requires true(Callable && callable)
  { static_cast<void>(callable); };

Remove unnecessary musing

Fix true edge case

Annotator tests (copied from format tests)

Yes, good point, I will do that

Add release note

Okay, I planned changes because I had more ambitious plans for fixing this, but those didn't work out, so instead I made a separate issue (https://github.com/llvm/llvm-project/issues/61469).
This patch in its current state is a simple fix, so it's probably fine as is

In D146140#4196886, @erichkeane wrote:

Ah, also, before commit, ensure that pre-commit CI passes.

Use dyn_cast_if_present, otherwise we segfault in some tests

utilize getCurFunction()

Slightly rewrite CheckForNonPrimary for slightly better clarity

Could you please clarify what you mean by "regressions" here? Isn't the behaviour of this syntax broken to begin with? It doesn't change anything about lambdas without requires-clauses

In D145642#4183539, @owenpan wrote:

Would this patch make the end result look worse without "making the requires clause formatting of lambdas
match the formatting for functions"? If yes, then we should not land it just yet.

Improve code flow in parseConstraintExpression