Thank you, this looks good to me.

I understand now that the intent behind the issue is that one may need to link libm in order to use frem, which was not obvious to me from the initial change.
I suggested an edit that conveys this more directly. Let me know what you think.

Sorry for the delay in replying to this. I don't think that such implementation details belong in the documentation of the operation.
Is there a particular problem that would have been avoided by adding in this information?

Glad you like it! I'd love to eventually get this added to existing workflows, either as a pre- or a post-merge check, but no plan has been concretised so far.
I am hoping to get some feedback from users of it before I put together a RFC---so please let me know if it turns out to be useful to you!

Thinking about this more, I think that you are right and the added complexity of not hardcoding the comment prefixes may not be worth it.
I decided to go with "decide we do care, and hardcode (any comment marker|start-of-line), dropping the optional-ness and language detection".

Remove comment prefix derivation and hardcode prefixes instead.

Thanks for the thorough review, Sam, I think this is much better!

Address review comments.

Abandoned in favour of https://reviews.llvm.org/D140846.

Thanks for the review, @tra and @herhut!

Address comments requested by reviewers.

The proposed fix is at https://reviews.llvm.org/D140846. Thanks again for pointing me there!

Fix vertical alignment.

Thanks for pointing this out, I agree that this would be a better fix.
I will work on a fix there.

The fix above corrects an edge case in which the result of the remainder operation is NaN when the second argument is inf.
This was detected in TensorFlow (https://github.com/tensorflow/tensorflow/issues/58369). (Sorry for the delay, I had to fix a problem with my bug repro :-))

Fix clang format errors.