detecting dialect in Racket

Question

I would like to write code that is correct for both untyped Racket and typed/racket...

For now, I've come up with the following not-so-satisfying principle: I insert a dummy syntax definition for ':' at the beginning of the file, so that the annotations are skipped when using untyped racket. And I comment this definition when using typed/racket.

E.g:

#lang racket
; comment the following line for typed/racket
(define-syntax : (syntax-rules () ((_ id type) (void))))
(: fact (-> Integer Integer))

; the rest of the file is common to both racket and typed/racket
(define (fact n) (if (zero? n) 1 (* n (fact (sub1 n)))))

and

#lang typed/racket
; comment the following line for typed/racket
;(define-syntax : (syntax-rules () ((_ id type) (void))))
(: fact (-> Integer Integer))

; the rest of the file is common to both racket and typed/racket
(define (fact n) (if (zero? n) 1 (* n (fact (sub1 n)))))

This way I just need to remove/insert a ';' to switch between racket and typed/racket...

But is there a way to have a code that doesn't need any change to run in racket and typed/racket? I didn't find how to programmatically detect if I'm in racket or typed/racket... Then I guess I would also have to find a way to conditionally define ':' at the top-level... All this doesn't seem the way to go, so, is there a better way ?

Alexis King Alexis King · Accepted Answer · 2018-01-19T17:57:49

Generally, if you want code to work with both #lang racket and #lang typed/racket, just write your code in #lang typed/racket. When you require a module written in #lang typed/racket from a module written in #lang racket, contracts will automatically be inserted between the two modules to enforce the types. You don’t have to do anything special to require a typed module from an untyped one, so from the user’s perspective, they don’t have to care.

The one area where you might need to worry about being in a typed or untyped context is when writing macros. Often, you can expand to the same code in both situations, but sometimes, you have to do things differently depending on whether or not the target code is typed. For that, you can use define-typed/untyped-identifier, which allows specifying two different forms to be used in different contexts.

If you really need to detect whether or not the current expansion context is typed, you can use the large hammer syntax-local-typed-context?. However, the documentation itself recommends avoiding its use:

This is the nuclear option, provided because it is sometimes, but rarely, useful. Avoid.

detecting dialect in Racket

1 Answers