Alex Knauth

Would a set of types like this make sense? ``` racket (Boxof/Write-Read w r) ; can write w, read r (Boxof/Write w) ; = (Boxof/Write-Read w Any) (Boxof/Read r) ;...

Right now, `Boxof` is a polymorphic type, not a type constructor syntax (like `->` for example). If it were a syntax type constructor, there would have to be a special...

Ok, I just renamed `Read-Only-Boxof` to `Boxof/Read` and added `Boxof/Write` as well. Read operations such as `unbox` take `(Boxof/Read a)` as arguments, and write operations such as `set-box!` take `(Boxof/Write...

What do you think? Is it worth another special case in `parse-type` to make `Boxof` take an optional argument, or should I leave it as is?

I need to rebase again, and then I need to implement the new `parse-type` case for `Box` for the optional argument. I think I'll re-visit this that after the current...

I'll pick it up again starting this Thursday.

Who should I ask about optimizing `box/sc` contracts with contravariant and covariant write and read fields? (Edit: Resolved)

That makes sense. However, the reason I asked is that without me adding anything specifically for `box/sc` optimizations, the optimizer still seems to be trying to optimize parts of them...

Is the "actual" correct for all 10 of those failures, for both "positive" and "negative" as trusted sides?

The only remaining test failures are these, where the "expected" and "actual" look the same visually printed out, but the test thinks they're different. The test uses `(and (subtype A...