Thanks. Yes, I realized that the two gave the same answer, but for 
debugging I tried to see if it was different. I think I now see where I am 
thinking about this in the wrong way. I thought `(U A B)` means 'it is 
either of type A or of type B', with the implication that it picks the more 
stringent one when possible. When there are two separate types A and B, 
this is innocuous enough, I think. What you are saying is that when typed 
racket sees `(U response Any)`, it transforms it (via some macro applied 
before doing type checking?) into `Any`. I thought it would leave `(U 
response Any)` around and always check against both, which is why I 
switched the order to see if that happened. 

A trivial and obvious answer to something that stumped me for quite a 
while. Good illustration of 

“It's not what we don't know that hurts. It's what we know that ain't so.”

Cheers,
Marc

On Wednesday, December 11, 2019 at 4:27:03 PM UTC+1, Sam Tobin-Hochstadt 
wrote:
>
> First, (U Any response) is the same as (U response Any) which is the 
> same as Any -- Any includes all other types and thus includes 
> response. 
>
> Second, f4 really is breaking the contract -- the contract Any turns 
> into says: don't try to pass through any "interesting" values, or if 
> you do, the other side isn't allowed to look at them. You passed a 
> "response" which is an interesting value (not a number or suchlike) 
> and the web server actually did something with it. The REPL doesn't 
> give you this error because in a typed module the REPL is also typed. 
>
> The problem is fundamentally that when you using Any in an annotation, 
> you're explicitly asking Typed Racket to throw away information. 
> Unfortunately, when you throw away that information for something that 
> you provide to untyped parts of your program, you've thrown away the 
> information Typed Racket needed to generate a more useful contract. So 
> instead, Typed Racket generates the best contract it can, which is as 
> described above. 
>
> Sam 
>
> On Wed, Dec 11, 2019 at 8:31 AM Marc Kaufmann <marc.ka...@gmail.com 
> <javascript:>> wrote: 
> > 
> > Hello, 
> > 
> > I have one file called `type-test.rkt` with the following (notice that I 
> discovered that there is a typed version of the web-server/http module, 
> which solves another of my issues): 
> > 
> > ``` 
> > #lang typed/racket 
> > 
> > (require (only-in typed/web-server/http response/xexpr response)) 
> > 
> > (provide f1 f2 f3 f4) 
> > 
> > (: f1 (-> response)) 
> > (define (f1) 
> >   (define x '(body (h1 "Try it"))) 
> >   (: resp response) 
> >   (define resp (response/xexpr x)) 
> >   resp) 
> > 
> > (: f2 (-> (U response Any))) 
> > (define (f2) 
> >   (define x '(body (h1 "Try it"))) 
> >   (: resp response) 
> >   (define resp (response/xexpr x)) 
> >   resp) 
> > 
> > (: f3 (-> (U response Number))) 
> > (define (f3) 
> >   (define x '(body (h1 "Try it"))) 
> >   (: resp response) 
> >   (define resp (response/xexpr x)) 
> >   resp) 
> > 
> > (: f4 (-> (U Any response))) 
> > (define (f4) 
> >   (define x '(body (h1 "Try it"))) 
> >   (: resp response) 
> >   (define resp (response/xexpr x)) 
> >   resp) 
> > ``` 
> > 
> > Then I have another *untyped* file for a servlet: 
> > 
> > ``` 
> > #lang racket 
> > 
> > (require "type-test.rkt" 
> >          web-server/servlet 
> >          web-server/servlet-env) 
> > 
> > (define (start req) 
> >   (f1) 
> >   ; (f2) 
> >   ; (f3) 
> >   ; (f4) 
> >   ) 
> > 
> > (serve/servlet start 
> >                #:servlet-regexp #rx"" 
> >                #:launch-browser? #false 
> >                #:port 8080) 
> > ``` 
> > 
> > Notice that I am telling all the f* functions that `resp` is of type 
> `response`. Yet, when I run the server with `start` using `f1` through `f4` 
> I get the following results: 
> > 
> > (f1): All good 
> > (f2): Error, see below. Unable to protect opaque value passed as `Any` 
> > (f3): All good 
> > (f4): Error, see below. 
> > 
> > The error is: 
> > 
> > ``` 
> > 
> > f4: broke its own contract 
> >   any-wrap/c: Unable to protect opaque value passed as `Any` 
> >   value: #<response> 
> >   in: the range of 
> >       (-> Any) 
> > 
> > ``` 
> > 
> > First, I couldn't figure out how to replicate this in the REPL, and had 
> to use the server to get the result. But I was able to figure out at the 
> REPL that (f2) and (f4) return something of type `Any`, for some unknown 
> reason. Clearly TR is smart enough to figure out that `resp` is not a 
> Number, but a response, but then when I allow it to return both a type 
> `response` and `Any`, it says that it's return value is `Any`. Why? Every 
> function that can take `Any` can take `response`, and every function that 
> expects response is now going to blow up (as now happens). 
> > 
> > At the REPL, I didn't manage to get this behavior, probably because 
> `serve/servlet` has contracts around it's arguments. Thus: 
> > 
> > - I pass all the typed racket tests 
> > - I nonetheless return something of type `Any` (which is really 
> guaranteed to be of type response, and I am already annotating, so this 
> surprises me) 
> > - When this hits the contract, it complains and tells me that f4 broke 
> its own contract, which seems false, but this may be one of those 'Contract 
> blaming is hard' moments 
> > 
> > Long story short: Why do (f2) and (f4) return something of type `Any` 
> rather than `response`? What is the logic for doing this and what use case 
> am I missing where this is a feature (or maybe it's just hard to get right, 
> but this seems vastly easier than other things Typed Racket does). 
> > 
> > And yes, I should probably just use typed/web-server/servlet and friends 
> (not sure if everything is covered, but probably) - but I discovered the 
> typed/** modules only while trying to fix this. And the issue will come up 
> with other non-typed modules or when/if I try to created typed versions of 
> some module. 
> > 
> > Cheers, 
> > Marc 
> > 
> > -- 
> > You received this message because you are subscribed to the Google 
> Groups "Racket Users" group. 
> > To unsubscribe from this group and stop receiving emails from it, send 
> an email to racket...@googlegroups.com <javascript:>. 
> > To view this discussion on the web visit 
> https://groups.google.com/d/msgid/racket-users/777af0ca-0b1f-474a-bd9f-8a9786e2a5c1%40googlegroups.com.
>  
>
>

-- 
You received this message because you are subscribed to the Google Groups 
"Racket Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to racket-users+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/racket-users/3dc7a351-ea55-4f70-a462-176559ed7183%40googlegroups.com.

Reply via email to