On Mon, Feb 16, 2015 at 3:16 PM, William ML Leslie <
[email protected]> wrote:

> On 17 February 2015 at 06:54, Matt Oliveri <[email protected]> wrote:
>
>> > What I was trying to say is that when all functions are arity-1
>> > the compiler never needs to insert lambda wrappers to turn curried
>> > application into intermediate function calls that accumulate the
>> arguments.
>>
>> Oh! OK, I understand now. So in other words, because curried
>> applications are only allowed on functions that are natively curried,
>> the compiler doesn't need to generate lambdas to curry them.
>>
>
> ​I think I missed something.  Wasn't the issue that the lambdas introduced
> allocation?  Natively curried functions typically do the same, right?  So
> what problem is this solving?
>

If all procedures have arity 1,  then curry-style application is only legal
where a procedure returns a lambda explicitly.

The key word here being "explicitly".


shap
_______________________________________________
bitc-dev mailing list
[email protected]
http://www.coyotos.org/mailman/listinfo/bitc-dev

Reply via email to