Mark Engelberg writes:
> It is my understanding that Typed Racket programs do not run any
> faster than their dynamically-typed counterparts, and in fact
> commonly run slower because there are a lot of additional runtime
> checks that must be inserted to handle various types of
On Fri, Jul 16, 2010 at 1:23 AM, ntu...@googlemail.com <
ntu...@googlemail.com> wrote:
> On Jul 15, 8:16 pm, Tomi Neste wrote:
> > But I don't think it would be easy to make it work with Clojure,
> > given how polymorphic and dynamic the language is (IMHO Scheme is not too
> > far from ML when it
A big part of it is that Clojure, being based on the JVM, has to deal with
inheritance and polymorphism in the core type system. An algebraic type system
like that of Haskell, OCaml, Scheme doesn't have to deal with inheritance.
There is parametric polymorphism, but that's based on type classes,
On Jul 15, 8:16 pm, Tomi Neste wrote:
> But I don't think it would be easy to make it work with Clojure,
> given how polymorphic and dynamic the language is (IMHO Scheme is not too
> far from ML when it comes to type systems).
Please expand.
- nt
--
You received this message because you are su
On Thu, Jul 15, 2010 at 7:16 PM, Tomi Neste wrote:
> www.ccs.neu.edu/*scheme*/pubs/popl08-thf.pdf
Thanks for the link, I missed that. Looks very interesting.
The filter one is more impressive, because it means that there is no
cheating involve, that would not be first-class.
But you're right C
would be hundreds of
> special cases)
>
Actually the previous example typechecks just fine with Typed Racket. Same
with something like
(apply + (filter number? '(1 foo "bar" 2)))
Exactly these kind of cases make Typed Rackets occurrence typing system
interesting ( www.ccs.neu.
It's a hard problem. Most type-systems forbid the kind of program people
write in Clojure
For example (if (keyword? x) using the fact that x is a keyword...
would not be typable in many languages. (Or there would be hundreds of
special cases)
There is a category of type systems, that can hand
:
> 2010/7/15 Mark Engelberg
>
> > The nice thing about Racket is the way you can write different parts
> > of your program in different Racket languages. So you can write some
> > pieces in Typed Racket, and others in Lazy Racket, and others in
> > standard Racket.
>
2010/7/15 Mark Engelberg
> The nice thing about Racket is the way you can write different parts
> of your program in different Racket languages. So you can write some
> pieces in Typed Racket, and others in Lazy Racket, and others in
> standard Racket.
>
> It is my unders
On Wed, Jul 14, 2010 at 3:49 PM, Mark Engelberg
wrote:
> purely about safety, not about speed. My guess is that the Clojure
> community would have little interest in any version of static typing
> that did not provide performance benefits.
check out the approach Dialyzer takes for Erlang. would
The nice thing about Racket is the way you can write different parts
of your program in different Racket languages. So you can write some
pieces in Typed Racket, and others in Lazy Racket, and others in
standard Racket.
It is my understanding that Typed Racket programs do not run any
faster than
For Racket (formely PLT Scheme), there exists a dialect called "Typed
Racket" [1], which allows for static type checking. I wonder if it is
feasible to port the typechecker to Clojure? Any ideas?
- nt
-
[1] http://docs.racket-lang.org/ts-guide/index.html
--
You received th
12 matches
Mail list logo
