On Wed, Feb 25, 2015 at 2:02 AM, Keean Schupke <[email protected]> wrote:
> 2) I still don't understand why you think these other methods of
> specialisation are not useable when you have a subtyping relationship in the
> type system?

The reason my proposal doesn't use subtyping is not because I think
subtyping is not flexible enough. It's because I think it's too
flexible! It allows things that I don't see how to implement without
implicit allocations. Please don't ask me to repeat the example that
we discussed all day.

> If I have a relation S <: T all it says is I am free to use an S anywhere I
> might use a T,  whether the reason I am able to do that is rewriting
> brackets in applications or any other mechanism.
>
> Keean.
>
> On 25 Feb 2015 00:19, "Matt Oliveri" <[email protected]> wrote:
>>
>> Here's what I think the situation is in this discussion:
>> - Matt (me), Shap, and William are skeptical that subtyping can be
>> implemented without introducing allocations.
>> - Keean doesn't see how subtyping would make it any harder to avoid
>> introducing allocations.
>> - Pal has joined in, but I'm not sure what his take is.
>> - Matt Rice keeps bringing up SML records, and I'm not sure why. (Sorry,
>> Matt.)
>> - Everyone else has been silent for a while.
>>
>> Again this is what I think is the situation. It's how it seems to me.
>> Please correct me if I'm wrong.
>>
>> But if I'm right, can we stop talking about subtyping for a while, and
>> tie up the other loose ends? The ones that are on my mind are
>> 1) deep vs. shallow arity variables vs. type constraints
>> 2) application-driven specialization without using subtyping
_______________________________________________
bitc-dev mailing list
[email protected]
http://www.coyotos.org/mailman/listinfo/bitc-dev

Reply via email to