on learning ocaml

a couple of questions:

1)
other than the obvious steps one can take like reading sources, the online
tutorials, and the online book - what other resources to people reccomend as
being particularly useful for learning/using ocaml? i'm coming from a very
strong OO background and think that the largest hurdle will be understanding
and embracing the functional paradigm so suggestions to that effect would be most
useful.

2)
how stable has ocaml been? i work in a research lab so i have the luxury of
using alpha software and development tools; however, the curse of successful
research is that some management type insists it get thrown straight into
production so __reasonable__ stability (of the language syntax and compiler
toolchain) is of importance, albeit secondary.

3) what scientifc libraries are used/most popular for ocaml? and scientific
image processing libraries out there? (by scientific image processing i
mean algorithims capable of handling raw binary, geotiff, hdl, etc. with
file sizes > 2gb in an efficient manner).

kind regards.

-a
--
================================================== =============================
| EMAIL :: Ara [dot] T [dot] Howard [at] noaa [dot] gov
| PHONE :: 303.497.6469
| A flower falls, even though we love it;
| and a weed grows, even though we do not love it.
| --Dogen
================================================== =============================

On Mon, 19 Jul 2004, Brandon J. Van Every wrote:

> Ara.T.Howard wrote:
>>
>> 1)
>> other than the obvious steps one can take like reading sources,
>> the online tutorials, and the online book - what other resources
>> to people reccomend as being particularly useful for
>> learning/using ocaml? i'm coming from a very strong OO background
>> and think that the largest hurdle will be understanding and
>> embracing the functional paradigm so suggestions to that effect
>> would be most useful.
>
> Start debates with people on comp.lang.functional about various issues. Or,
> if you don't want to, read the archives about questions I've recently asked.
> "FP tutorial for OO programmers?"
> http://groups.google.com/groups?selm...0uni-berlin.de
> Warning, this branches off into nearly 400 posts! I've lost track of what a
> lot of people have said, and it hasn't always been terribly relevant or
> clear, but sometimes someone says something succinct that helps my
> understanding. If you don't mind following long Usenet debates, you could
> either read the archives gradually at your leisure, or post your own 'new'
> questions.
>
> There is also the ocaml-beginners mailing list.
> yahoo.com/group/ocaml_beginners/" target="_blank">http://groups.yahoo.com/group/ocaml_beginners/
> They're having a good discussion of Functors right now. Although actually,
> the best gloss I've heard of Functors is "they're like Templates in C++,
> only more powerful and general purpose." For instance you can parameterize
> them with Structures, which in turn are fairly arbitrary collections of
> stuff.

good advice - i read newgroups alot, but seldom retrospectively. can i ask
what type of projects you've been using ocaml for?

kind regards.

-a
--
================================================== =============================
| EMAIL :: Ara [dot] T [dot] Howard [at] noaa [dot] gov
| PHONE :: 303.497.6469
| A flower falls, even though we love it;
| and a weed grows, even though we do not love it.
| --Dogen
================================================== =============================

Ara.T.Howard wrote:
> On Mon, 19 Jul 2004, Brandon J. Van Every wrote:
> [about perusing newsgroup archives]
>
> good advice - i read newgroups alot, but seldom retrospectively. can
> i ask what type of projects you've been using ocaml for?

I will be using OCaml for 3D graphics, AI, and game development. I have a
lot of low level performance issues to resolve. 3D and AI have insatiable
appetites.

I haven't written a line of OCaml yet. :-) I take a managerial,
design-oriented, "top down" approach to programming, where actually doing
the programming is regarded as an implementation detail. My priorities when
evaluating a language are:

1) is there an extant community?
2) does the community have long term growth potential, i.e. could I make
money with it someday?
3) how does the community respond to my noise about marketing and growth
issues?
4) are there extant libraries that do what I need?
5) have people proven that they can do something useful in my problem domain
with the language?
6) what are the syntax and semantics of the language?
7) what does the low level C Foreign Function Interface look like?
8) what program design do I need to tackle my problems?

Only once I've run through all of these things do I start coding. I move
from strategy to tactics. That's 'cuz I've lost an awful lot of money on
tactics in the past. Bobbing and weaving and code monkeying really doesn't
matter if your whole plan is strategically DOA.

I've been looking at a lot of languages in the past year, running through
this drill. Previously I was pursuing a Python strategy, but it just didn't
have the performance I was looking for. In the beginning of April 2004 I
became more seriously interested in OCaml. It was previously on my radar
because of the Language Shootouts, the ICFP, and available SWIG support.
So, I've really only been working through OCaml issues for about 3 months.

I made it all the way to item (7) with OCaml, then I started getting cold
feet about its C FFI. I started looking around for alternatives in the ML
family. SML/NJ has a more interesting C FFI, called "No Longer Foreign
Function Interface." However, the SML/NJ community is moribund on points
(1) (2) (3). Also it turns out that SML/NJ doesn't support 32-bit floats
like I thought it did, it just offers convenient conversion to 64-bit
floats. So I started looking at OCaml again, because clearly their
community is a big win.

Recently I've been wrapping my head around Bigarray, and I've realized it's
the best way to talk to low level C routines. Bigarray has a lot of support
for C types, so long as the types are homogeneous. Homogeneity is a
restriction, but a surmountable one. So, I'm working through point (8) and
will probably start coding this week.

My provisional strategy is to write a small prototype game in OCaml, to
demonstrate proof of concept. I expect it will take me until early winter.
If it works, I'll keep going with OCaml. If I encounter serious problems,
I'll bail for Scheme. But I have this nice ML Seattle group started now, so
I hope it doesn't come to that.


--
Cheers, www.indiegamedesign.com
Brandon Van Every Seattle, WA

Taking risk where others will not.

Hi Ara,

To answer another of your questions:

In article <cdgnup$e75$1@wolfberry.srv.cs.cmu.edu>, Ara.T.Howard wrote:
> 2) how stable has ocaml been? i work in a research lab so i have the
> luxury of using alpha software and development tools; however, the curse
> of successful research is that some management type insists it get thrown
> straight into production so __reasonable__ stability (of the language
> syntax and compiler toolchain) is of importance, albeit secondary.

While i am not an O'Caml developer, i would say the language is reasonably
stable, by your definition. The developers recognize that for the language
to make it both industrially and academically (think textbooks), new
releases must remain syntactically and semantically compatible with
previous releases. Every now and then, the Right Thing[tm] to do involves
breaking backward compatibility, but typically before making any such
changes, the developers do an informal survey via the caml-list mailing
list to see if anyone currently depends upon the original behavior.

Hope this helps.

cheers,
William

On Thu, 22 Jul 2004, William Lovas wrote:

> Hi Ara,
>
> To answer another of your questions:
>
> In article <cdgnup$e75$1@wolfberry.srv.cs.cmu.edu>, Ara.T.Howard wrote:
>> 2) how stable has ocaml been? i work in a research lab so i have the
>> luxury of using alpha software and development tools; however, the curse
>> of successful research is that some management type insists it get thrown
>> straight into production so __reasonable__ stability (of the language
>> syntax and compiler toolchain) is of importance, albeit secondary.
>
> While i am not an O'Caml developer, i would say the language is reasonably
> stable, by your definition. The developers recognize that for the language
> to make it both industrially and academically (think textbooks), new
> releases must remain syntactically and semantically compatible with
> previous releases. Every now and then, the Right Thing[tm] to do involves
> breaking backward compatibility, but typically before making any such
> changes, the developers do an informal survey via the caml-list mailing
> list to see if anyone currently depends upon the original behavior.
>
> Hope this helps.
>
> cheers,
> William

i'd hoped as much. thanks.

-a
--
================================================== =============================
| EMAIL :: Ara [dot] T [dot] Howard [at] noaa [dot] gov
| PHONE :: 303.497.6469
| A flower falls, even though we love it;
| and a weed grows, even though we do not love it.
| --Dogen
================================================== =============================