Prev: MIPS Recursion Help
Next: * Program ALPHATEST (FORTRAN 77)
From: Jon Harrop on 1 Oct 2009 15:42
Dave Searles wrote:
> IFractional and BigRational? One's an interface and the other's the
> implementation.

I think he meant BigNum and BigRational. I just checked and they are just
aliases of the same type.

--
Dr Jon D Harrop, Flying Frog Consultancy Ltd.
http://www.ffconsultancy.com/?u
From: Jon Harrop on 1 Oct 2009 16:32
Dave Searles wrote:
> Jon Harrop wrote:
>> Lisp lacks static typing, is slow to run, impossible to parallelize (due
>> to the lack of any Lisp implementations with concurrent garbage
>> collectors, even among commercial offerings), comparatively
>> uninteroperable and has no libraries comparable to WPF, XNA, LINQ and the
>> TPL.
>
> Bullshit. One Lisp, Clojure...

I was talking about Common Lisp.

> has a concurrent GC,

Not by default.

> specific support for parallelism,

Anything comparable to the lock-free work stealing concurrent deques of the
Task Parallel Library?

> and is interoperable with the huge existing body of Java
> libraries and code.

Anything comparable to the robustly-deployable hardware accelerated 2D and
3D vector graphics of Windows Presentation Foundation?

> Furthermore, a version has been made for the .NET
> runtime which has the same concurrency support (modulo concurrent GC in
> the .NET runtime) and access to the same libraries your precious F# has
> access to.

They only just fixed value types in Clojure yesterday. You can hardly
compare that to something that Microsoft are going to release as
fully-supported production-quality software in a few weeks...

--
Dr Jon D Harrop, Flying Frog Consultancy Ltd.
http://www.ffconsultancy.com/?u
From: Mensanator on 1 Oct 2009 18:25
On Oct 1, 2:42 pm, Jon Harrop <j...(a)ffconsultancy.com> wrote:
> Dave Searles wrote:
> > IFractional and BigRational? One's an interface and the other's the
> > implementation.
>
> I think he meant BigNum and BigRational. I just checked and they are just
> aliases of the same type.

Perhaps they are aliases now, but they were distinct types originally.

I know that because somewhere in that esoteric documentation it
describes the two types, how one is more memory efficient, but does
not implement the same methods as the inefficient version.

Naturally, if you need such functionality, you MUST use the
inefficient version since no type conversion functions were
implemented. If I use the efficient version, then I lose
functionality.

That's one of the reasons I say F# is stupid.

Maybe they have corrected it by now, so there is one rational type
that is efficient and contains the complete set of methods.

This new release you speak of, what version is it? Or what version
will it be if they haven't released it yet?

Will there still be a free downloadable version or will I have to
buy the Visual Design Studio?

>
> --
> Dr Jon D Harrop, Flying Frog Consultancy Ltd.http://www.ffconsultancy.com/?u

From: Dave Searles on 2 Oct 2009 09:37
Jon Harrop wrote:
> Dave Searles wrote:
>> Jon Harrop wrote:
>>> Lisp lacks static typing, is slow to run, impossible to parallelize (due
>>> to the lack of any Lisp implementations with concurrent garbage
>>> collectors, even among commercial offerings), comparatively
>>> uninteroperable and has no libraries comparable to WPF, XNA, LINQ and the
>>> TPL.
>> Bullshit. One Lisp, Clojure...
>
> I was talking about Common Lisp.

No, you weren't. You simply said "Lisp lacks ...", not "Common Lisp
lacks ...".

>> has a concurrent GC,
>
> Not by default.

It has a concurrent GC.

>> specific support for parallelism,
>
> Anything comparable to the lock-free work stealing concurrent deques of the
> Task Parallel Library?

Lock-free futures do that, and there's a lock-free software
transactional memory for you, too.

>> and is interoperable with the huge existing body of Java
>> libraries and code.
>
> Anything comparable to the robustly-deployable hardware accelerated 2D and
> 3D vector graphics of Windows Presentation Foundation?

Java2D and Java3D are accelerated on most platforms, including Windows.

>> Furthermore, a version has been made for the .NET
>> runtime which has the same concurrency support (modulo concurrent GC in
>> the .NET runtime) and access to the same libraries your precious F# has
>> access to.
>
> [personal attack deleted]

Wrong.
From: Jon Harrop on 2 Oct 2009 21:02
Dave Searles wrote:
> Jon Harrop wrote:
>> Anything comparable to the lock-free work stealing concurrent deques of
>> the Task Parallel Library?
>
> Lock-free futures do that...

No, that is a phrase you just invented.

Wait-free work-stealing deques are the state-of-the-art solution to
task-based parallelism and are seen in solutions like Cilk and Microsoft's
Task Parallel Library (part on .NET 4).

Clojure doesn't have them.

>>> and is interoperable with the huge existing body of Java
>>> libraries and code.
>>
>> Anything comparable to the robustly-deployable hardware accelerated 2D
>> and 3D vector graphics of Windows Presentation Foundation?
>
> Java2D and Java3D are accelerated on most platforms, including Windows.

They weren't even close and they've been dead for some time now:

http://www.google.com/trends?q=java2d%2Cjava3d%2Cwpf&ctab=0&geo=all&date=all&sort=0

Clojure has nothing comparable to WPF (or XNA, or the TPL or ...) because
the JVM has nothing comparable. Indeed, in many cases the JVM is
fundamentally incapable of expressing such solutions: it doesn't even have
value types and tail calls.

>>> Furthermore, a version has been made for the .NET
>>> runtime which has the same concurrency support (modulo concurrent GC in
>>> the .NET runtime) and access to the same libraries your precious F# has
>>> access to.
>>
>> They only just fixed value types in Clojure yesterday. You can hardly
>> compare that to something that Microsoft are going to release as
>> fully-supported production-quality software in a few weeks...
>
> Wrong.

Not according to David Miller (the author of ClojureCLR) as of two days ago:

"The handling of non-primitive value types by the compiler still has
some problems." - David Miller, 30th September 2009

http://www.mail-archive.com/clojure%40googlegroups.com/msg18886.html

Microsoft have ten people working full-time on F# and have been working on
its foundation since 2001. There are already thousands of people
programming in F# in industry. You are comparing that to something that one
guy alpha released seven months ago that has virtually no users.

--
Dr Jon D Harrop, Flying Frog Consultancy Ltd.
http://www.ffconsultancy.com/?u
First  |  Prev  |  Next  |  Last
Pages: 1 2 3 4 5
Prev: MIPS Recursion Help
Next: * Program ALPHATEST (FORTRAN 77)