[Agda] Re: Agda FFI bindings

[Daniel Peebles]

Maybe there could be? You can already postulate obvious lies, so it seems
like it wouldn't be too unreasonable for a postulate to claim to be a
constructor too?

On Tue, Oct 5, 2010 at 3:22 AM, Alan Jeffrey <ajeffrey at bell-labs.com> wrote:

> On 10/04/2010 05:10 PM, Nils Anders Danielsson wrote:
>
>> On 2010-10-04 22:57, Alan Jeffrey wrote:
>>
>>> Yes, that I get, I just don't understand why not just include a
>>> postulate inf : forall {A} ->  \inf(IO A) ->  (IO A), and get rid of the
>>> distinction between IO.Primitive.IO and IO.IO.
>>>
>>
>> Because Agda uses guarded corecursion, and postulates are not treated as
>> constructors.
>>
>
> Urg.  Good point.  Using a coninductive datatype with a run function allows
> the --no-termination-check option to be local to the IO module, whereas
> there's no similar "honest I'm a constructor" option.
>
> A.
>
> _______________________________________________
> Agda mailing list
> Agda at lists.chalmers.se
> https://lists.chalmers.se/mailman/listinfo/agda
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.chalmers.se/pipermail/agda/attachments/20101005/91e93d73/attachment.html