There are two views in the exception world:
-- annotate the return type with all possible exceptions
(Java does this and is the most popular language that does so)
-- consider exception to be implicit effects
(ML created and pushed this variant for three decades)
PLT Scheme went with the second option (as an untyped language) from the
beginning. Typed Racket {Scheme} mirrors this decision.
Since the fail-thunk isn't expected to return anything, I would assume that
(almost) all calls to file-or-directory-modify-seconds get away with
(-> Void)
as the type for the fail thunk.
Other than that, the function can take 1, 2, or 3 arguments I would expect Sam
to use some case-lambda type for the function overall.
-- Matthias
On Sep 5, 2010, at 9:59 AM, Shriram Krishnamurthi wrote:
> Can you tell us (om the list) what the true type is? I'm sure I'm not
> the only one curious as to precisely what it is and how you would
> write it w/ the exception. Thanks.
>
> Shriram
>
> On Sun, Sep 5, 2010 at 8:57 AM, Sam Tobin-Hochstadt <sa...@ccs.neu.edu> wrote:
>> On Sun, Sep 5, 2010 at 3:06 AM, Hari Prashanth <krh...@ccs.neu.edu> wrote:
>>> Is this what you are looking for?
>>>
>>> (require/typed racket/base
>>> [
>>> (String (Option Integer) (-> exn:fail:filesystem) -> (U
>>> Integer Void))])
>>
>> Hari is correct that you can use `require/typed' on `racket/base'.
>> However, that type isn't quite right - it shouldn't *return*
>> `exn:fail:filesystem' - that is an exception that might be raised.
>>
>> The true type is more complex; I'll add it to the base environment.
>> --
>> sam th
>> sa...@ccs.neu.edu
>> _________________________________________________
>> For list-related administrative tasks:
>> http://lists.racket-lang.org/listinfo/dev
>>
> _________________________________________________
> For list-related administrative tasks:
> http://lists.racket-lang.org/listinfo/dev
_________________________________________________
For list-related administrative tasks:
http://lists.racket-lang.org/listinfo/dev
