Re: [Caml-list] typing problem with sexplib and mutually recursive polymorphic types

[Yoann Padioleau <padator@wanadoo.fr>]

Yitzhak Mandelbaum <yitzhakm@CS.Princeton.EDU> writes:

> While I'm not sure it will help you in this case (because the code is
> generated), 

The code is generated, but I am ok with modifying it. I use 
typeconv/sexplib/binprot and camlp4 just as a way to produce a first draft
and then I refine the result. I prefer to see the generated code
and have concrete files (and in this case in fact I have no
choice as the the default generated code is not good enough).

> I should note that type *inference* for polymorphic
> recursion is undecidable, but type *checking* is not. Moreover, its
> pretty easy to do in OCaml (if a bit clunky).
>
> An earlier discussion on this list covered a number of ways to do it.
> Here's one:
>
> http://groups.google.com/group/fa.caml/msg/d85e3f62dd52d365

Thanks. I tried the {f: 'a. 'a } trick and it works on my
simplified example but I failed using it on my original file. Never mind,
I artificially broke the mutually recursive dependency
and now it works. 

Thanks.

>
> Cheers,
> Yitzhak
>
>
> On Mar 11, 2009, at 10:20 AM, Markus Mottl wrote:
>
>> On Wed, Mar 11, 2009 at 02:11, yoann padioleau <padator@wanadoo.fr>
>> wrote:
>>> Yes on this example, but what if I can't move it out because
>>> the types really need to be mutually recursive ?
>>
>> I'm afraid, I don't see any straightforward way of achieving this
>> goal.  If the above were allowed in a truly recursive way, you'd
>> essentially end up with polymorphic recursion, which is undecidable.
>>
>> Regards,
>> Markus
>>
>> -- 
>> Markus Mottl        http://www.ocaml.info
>> markus.mottl@gmail.com
>>
>> _______________________________________________
>> Caml-list mailing list. Subscription management:
>> http://yquem.inria.fr/cgi-bin/mailman/listinfo/caml-list
>> Archives: http://caml.inria.fr
>> Beginner's list: http://groups.yahoo.com/group/ocaml_beginners
>> Bug reports: http://caml.inria.fr/bin/caml-bugs
>
> -----------------------------
> Yitzhak Mandelbaum