Re: Traits seem to be working.

[Ori Bernstein <ori@xxxxxxxxxxxxxx>]

At the moment, no. Now that I think about it, this would also be equivalent
to shorthand for generic impls. In other words:

    trait clonable @a =
        dup = {v : @a
            -> a
        }
    ;;

would be equivalent to:

    trait clonable @a =
        dup : (v : @a -> @a)
    ;;

    /* applies to all types */
    impl clonable @a =
        dup = {v
            -> v
        }
    ;;
    
Which leads to questions on how to pick the /most/ specific
specialization. A nice can of worms there.

On Sat, 21 Feb 2015 14:56:35 -0600
Ryan Gonzalez <rymg19@xxxxxxxxx> wrote:

> On that note...
> 
> Is it possible to provide a default value for a function in a trait?
> 
> On Sat, Feb 21, 2015 at 12:28 PM, Ori Bernstein <ori@xxxxxxxxxxxxxx> wrote:
> 
> > Go me. I was too lazy to announce this when I got it working, but I
> > realized
> > last night that I should really send out a mail.
> >
> > As of a few weeks ago, I got traits working -- You can now define traits
> > for types, and expect them to actually compile and run. Traits are close
> > to concepts in C++, and similar to compile time only interfaces in certain
> > other languages. An example would be:
> >
> >     /*
> >     Define a hashable trait. This will allow any type that
> >     has an implementation of this trait to get hashed.
> >     */
> >     trait hashable @a =
> >         hash : (val : @a -> uint32)
> >     ;;
> >
> >     /* hashing implementation for integers */
> >     impl hash int =
> >         hash = {val
> >             -> val * Bigprime
> >         }
> >     ;;
> >
> >     /* hashing implementation for strings */
> >     impl hash byte[:] =
> >         hash = {str
> >             var h = 1
> >             for b in str
> >                 h = 33 * h * (b castto(int32))
> >             ;;
> >         }
> >     ;;
> >
> >     /* use the hashing */
> >     const main = {
> >         std.put("hash of 123: %i\n", hash(123))
> >         std.put("hash of "abc": %i\n", hash("abc"))
> >         /* this will error out, saying there's no trait implemented */
> >         /* std.put("hash of 'c': \n", hash('c')) */
> >     }
> >
> >
> > There are currently a few limitations on traits that I would like to
> > lift. The biggest one is that currently, it only works on concrete types.
> > You can't, for example, implement:
> >
> >      impl hash @a::(numeric,integral) = ... ;;
> >
> > and get a hash implementation for every numeric type.
> >
> > This feature probably means that the API of some libraries could probably
> > be adjusted.
> >
> >
> 
> 
> -- 
> Ryan
> If anybody ever asks me why I prefer C++ to C, my answer will be simple:
> "It's becauseslejfp23(@#Q*(E*EIdc-SEGFAULT. Wait, I don't think that was
> nul-terminated."
> Personal reality distortion fields are immune to contradictory evidence. -
> srean
> Check out my website: http://kirbyfan64.github.io/