Re: Dylan vs Julia

On Fri, 14 Feb 2014 17:04:09 -0600
mikel evins <[email protected]> wrote:

> As languages, Julia and Dylan are strikingly similar. The Julia
> maintainers have said that's because they have mostly the same
> influences, rather than because of any deliberate intention to
> imitate Dylan.

Yep, although it's hard to believe that Dylan was not considered
directly. :-)

> Their syntax is different, of course. Dylan's syntax is broadly
> Pascal-like; Julia's is broadly Matlab-like.

Considering that I was programming in (Turbo) Pascal long ago (it was my
3rd language after Basic & Fortran) and never used Matlab, Dylan has
advantage for me here. :-)

> Julia's definition includes primitives for parallel and distributed
> computations; Dylan's doesn't. For distributed and concurrent
> programming in Dylan you must rely on whatever libraries are
> available, or write one yourself.

If/when Dylan takes off, it will probably take care about it not to miss
multi-core capabilities.

> The type is called a parametric or parametrized type because part of
> its definition is a type parameter--in this case type T. Type T can
> be any type you like (but it has to be the same type in both x and
> y). That means that Point{T} is now actually a family of types like
> Point{Int64}, Point{Float64}, and Point{String}. Julia functions can
> be specialized on any of these types, or on Point{T}.

That's really cool feature.

> Dylan does not have this feature. On the other hand, Dylan does have
> limited types, which enable you to achieve many of the same effects.

I really have to take closer look at it 'cause it is something we'd
need.

> You can then specialize functions on <integer-vector>, achieving the
> same effect as using Julia's Vector{Integer}. Julia's parametric
> types are probably strictly more expressive, because they are
> generic; you can use them with arbitrary types. In Dylan you have
> support only for those types that the implementation supports as
> limited types, and you can make only those limitations that the
> implementation provides. On the other hand, if your use-case matches
> the features of your Dylan implementation then Julia's parametric
> types offer no unique advantage for that case.

Thank you. I have to study it more to get proper apprehension of those
features.


Sincerely,
Gour

-- 
An intelligent person does not take part in the sources of misery, 
which are due to contact with the material senses. O son of Kuntī, 
such pleasures have a beginning and an end, and so the wise man does 
not delight in them.

http://www.atmarama.net | Hlapicina (Croatia) | GPG: 52B5C810
_______________________________________________
hackers mailing list
[email protected]
https://lists.opendylan.org/mailman/listinfo/hackers