Noah Ma

I'm not sure about the purpose of introducing `section` in `defmodule/incl`, but I tried deleting `#:style` in these two definitions to ensure that the section is valid: ``` @(define-syntax-rule @defmodule/incl[name...

By the way, can we consider adding `@section{Typed for the @racketfont{typed} collection}` and using ``` (subsection #:style '(hidden) (symbol->string 'name)) ``` instead of ``` (section #:style '(hidden toc-hidden unnumbered) (string-append...

### What version of Racket are you using? `8.0` ### What program did you run? ``` #lang typed/racket (: ls (Pair Number (Listof Number))) (define ls '(0 1 2 3...

I can see this difficulty. In fact, I thought about creating the PR when I discovered that there was a hidden bug in my code which should be thrown by...

I think it's safe to refine operators' return types. And the default return type of `for[*]/hash[eq/eqv]:` should be `#'(Immutable-HashTable Any Any)`. In this way, we can combine the definitions of...

Hi @samth, currently this PR only optimizes the return type, I think this change is safe. Will TR consider merging it?

> the changes to the type annotations in the tests worry me. I thought `(Immutable-HashTable A B)` is the subtype of `(HashTable A B)` and these changes should be safe,...

### What version of Racket are you using? `v8.3` ### What program did you run? ``` #lang typed/racket/base (define-type Non-Hash (Refine [t : Any] (! t HashTableTop))) (ann eof Non-Hash)...

I hope TR can be more flexible for composing functions (I'm interested in building `typed/qi`), but I can't think of a solution other than treating the function arguments as similar...

> A typed version of Qi would be very cool, but Qi doesn't change how functions work in Racket; it uses the underlying mechanism. Although TR doesn't regard function arguments...