Alexis King

> I do also see the paragraph in https://docs.racket-lang.org/reference/creatingunits.html that says "[…] a unit is prohibited syntactically from […] exporting an identifier that is imported." Why is this restriction in...

One thought I’ve had since writing this proposal is that there is a potentially simpler alternative, namely a dependent version of `unit/c`. Then I could write something like this: ```racket...

> The model of signatures with expressions as effectively containing an embedded unit seems quite attractive. It seems like it might be able to explain more features of units, also....

I have been working on a significantly reworked implementation of unit and signatures contracts, including support for `->i`-style “indy” blame semantics for `unit/c`, and I’ve run into a couple things...

> Reading through old discussions, I saw that we talked about allowing signatures to import signatures in #2604. Hah! I had completely forgotten I had opened that issue. I guess...

> This comment prompted me to wonder if the following program would work: unfortunately (but as I expected), it fails with the error `define-values/invoke-unit/infer: not a unit definition in: x@`...

Now that #4567 has landed as ae85ce6a5c2c43030372346b1f6f0f7d0cf7bf78, which includes the improvements to `unit/c` and `define-unit/contracted`, I think this issue is a significantly less pressing concern for me: using `unit/c` is...

Here’s a perhaps scarier example that also uses stateful rename transformers, though this one doesn’t break any internal invariants as far as I can tell, it just asks questions and...

I’m taking a look at this now, and I’m trying to wrap my head around it (and not really succeeding; my intuition for use-site scopes is not great). I don’t...

Yes, I think your explanation mostly makes sense to me. I believe I do understand the purpose of use-site scopes, and I think I have a reasonable intuition for them...