Cameron Moy

`impersonator-prop:application-mark` is one way of doing it. Instead, I'm using this feature of `impersonate-procedure`: > any number of repetitions of 'mark key val (i.e., three values), where the call proc...

Yes, that's exactly the reason. I figure most Racket programmers are familiar with parameters, but probably not continuation marks, so a parameter-based API makes sense to me. We could also...

You're right that the error message isn't great, but it's actually the same error that you get from `parameterize` itself (in terms of `extend-parameterization`): ```rkt > (parameterize ([#t #t]) #t)...

Ok, sounds good. [Here](https://github.com/racket/racket/pull/4361/files#diff-a1a913f821f46f019b1d8b7d56d3c1cbba19fef4f89a0a7e7b37151544ff3f06R760) is the `parameter?` check.

Funny, `raise-argument-error` supports specifying a positional argument in the error message, but not a keyword argument. So, I've just followed what `dynamic->*` does and made the expected string "parameter? in...

> that this would change any assumptions that currently hold about the way parameters can be manipulated by safe/public Racket. Indeed, as @rfindler said, this should not be the case....

@rfindler anything else that you want me to do before this PR can be merged?

Maybe I'm not understanding you correctly, but I think it's already like that. You have a `#:param` for each one. Here's one of the [unit tests](https://github.com/racket/racket/blob/48005791b38dbe8390d2bcb5c4329f01b77edc18/pkgs/racket-test/tests/racket/contract/arrow-i.rkt#L1310).

So I did the sanity checks you asked and other than the additional (empty) argument to `make-->i` there's no changes to the generated code. In my last push I changed...

Sounds good, I've updated the version number accordingly. Thanks for your help!