RE: [Sml-implementers] SML IDE as language promotion

[Nick B. <ni...@mi...>]

Strange, twisted versions of the ideas are indeed making it to
widespread adoption. Look at the latest version of C# - not just types
and garbage collection, but parametric polymorphism, anonymous
functions, etc. Even limited forms of type inference. Andrew Kennedy has
done various bits of classical functional programming in C#2.0 (parser
combinators, functional graphics) and whilst it's not pretty, it's
certainly possible. With Claudio Russo, he's now even showing how to do
GADTs, one of the latest things in the pure FP community, in C#.
Continuations (and the benefits of statelessness) are a hot topic in web
programming...

So we either=20
(a) declare victory
(b) keep trying to make the argument. This is harder than before since
instead of saying "we have types/gc/higher order functions and you
don't" we have to explain how we have the "right" versions of those
features, and how leaving some other features out makes those features
more usable.
(c) come up with something even better

I thing (b) becomes increasingly untenable with respect to "straight"
SML. There are some exciting developments (MLton, for example (insert
gripes about Windows support)) but the language has to evolve.=20

Poly/ML: www.polyml.org, though I can't get a response at this very
moment.

Anyway, I should go and do some (c). Anyone with substantive opinions on
such matters should write them up and submit them to the 2005 ML
Workshop:
http://ttic.uchicago.edu/~blume/ml05/.

  Nick

[No - I'm Spartacus. I was two years earlier than you on the Maths
Tripos and the Diploma.]

-----Original Message-----
From: nb...@ra... [mailto:nb...@ra...] On Behalf Of Nick
Barnes
Sent: 14 March 2005 13:35
To: Nick Benton
Cc: sml...@li...
Subject: Re: [Sml-implementers] SML IDE as language promotion=20

At 2005-03-14 11:23:37+0000, "Nick Benton" writes:

> The people who like SML just don't seem all that keen on writing
> applications other than theorem provers and SML compilers.

I'd be tempted, and would certainly try it, if it had anything like
the library support that Python has (almost all of it portably between
*nix and Windows, which is important for a lot of the work that we
do).  But I recognize that I'm in a minority.  Way back when, the
MLWorkers looked around for real-world ML use.  I remember using some
'real applications' (academic biochemistry computations, as I recall)
as test cases for the MLWorks profiler.

So has SML permanently missed this boat?  Is there a future for SML in
which it gets used for real programming?  Or at least for the ideas
which SML incorporates and represents?  SML.NET is an excellent step
in that direction (albeit single platform etc etc, insert gripes about
Microsoft here, I guess SML.NET/Mono will work one day).  I should get
around to installing the copy of Visual Studio .NET that Mike Smith
mailed to me....

What ever happened to Poly/ML?

[the other? the original?] Nick B