Being on sabbatical has given me a bit of experience with other systems and languages. Also, my kids are now old enough to “mess around” with programming. Learning from both of these, I’d like to hazard a bit of HtDP heresy: students should learn for i = 1 to 10 before they learn

1
2
3

(define (sum lon)
  (cond [(empty? lon) 0]
        [else (+ (first lon) (sum (rest lon)))]))

To many of you, this may seem obvious. I’m not writing to you. Or maybe you folks can just read along and nod sagely.

HtDP takes this small and very lovely thing—recursive traversals over inductively defined data—and shows how it covers a huge piece of real estate. Really, if students could just understand how to write this class of programs effectively, they would have a vastly easier time with much of the rest of their programming careers, to say nothing of the remainder of their undergraduate tenure. Throw a few twists in there—a bit of mutation for efficiency, some memoization, some dynamic programming—and you’re pretty much done with the programming part of your first four years.

The sad thing is that many, many students make it through an entire four-year curriculum without ever really figuring out how to write a simple recursive traversal of an inductively defined data structure. This makes professors sad.

Among the Very Simple applications of this nice idea is that of “indexes.” That is, the natural numbers can be regarded as an inductively defined set, where a natural number is either 0 or the successor of a natural number. This allows you to regard any kind of indexing loop as simply a special case of … a recursive traversal of an inductively defined data structure.

So here’s the problem: in September, you face a bunch of bright-eyed, enthusiastic, deeply forgiving first-year college students. And you give them the recursive traversal of the inductively defined data structure. A very small number of them get it, and they’re off to the races. The rest of them struggle, and struggle, and finally get their teammates to help them write the code, and really wish they’d taken some other class.

NB: the rest of this makes less sense… even to me. Not finished.

However, another big part of the problem is … well, monads are like burritos.

Let me take a step back.

The notion of repeated action is a visceral and easily-understood one. Here’s what I mean. “A human can multiply a pair of 32-bit integers in about a minute. A computer can multiply 32-bit integers at a rate of several billion per second, or about a hundred billion times as fast as a person.” That’s an easily-understood claim: we understand what it means to the same thing a whole bunch of times really fast.

So, when I write

for i=[1..100] multiply_two_numbers();

It’s pretty easy to understand that I’m doing something one hundred times.

Is this post a thinly disguised ripoff of Brian Anderson’s post about embedding Rust in Ruby? Why yes. Yes it is.

Okay, let me start with a little background. Rust is a magnificent language that comes from Mozilla; it’s targeted at programmers who want

• high and predictable performance,
• control over memory layout,
• good support for concurrency, and
• safety.

I think the Mozilla Research homepage is probably the best place to start learning about Rust.

To be honest, though, I’m probably flattering myself if I think that this blog post is being read by anyone who doesn’t already know lots about Rust.

One of the key requirements of a language like Rust is that it be embeddable; that is, it should be possible to call Rust code from another language just as it’s possible to call C code from another language.

This is now possible.

To illustrate this, Brian Anderson posted a lovely example of embedding Rust in Ruby. But of course, embedding Rust in Ruby is pretty much exactly the same as embedding Rust in any other language.

Say, for instance, Racket.

So, without further ado, here’s the setup. You just happen to have a small web app written in Racket that performs a Gaussian Blur. You decide to optimize the performance by porting your code to Rust. Then you want to plug your Rust code into your Racket application. Done! Here’s the github repo that contains all of the code.

Let’s see that again in slow motion.

First, here’s the gaussian blur function, written in Racket. We’re going to stick with a grayscale image. It works fine in color, but the code is just that much harder to read.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33

;; the gaussian filter used in the racket blur.
;; boosted center value by 1/1000 to make sure that whites stay white.
(define filter '[[0.011 0.084 0.011]
                 [0.084 0.620 0.084]
                 [0.011 0.084 0.011]])

;; racket-blur: blur the image using the gaussian filter
;; number number list-of-bytes -> vector-of-bytes
(define (racket-blur width height data)
  (define data-vec (list->vector data))
  ;; ij->offset : compute the offset of the pixel data within the buffer
  (define (ij->offset i j)
    (+ i (* j width)))
  (define bytes-len (* width height))
  (define new-bytes (make-vector bytes-len 0))
  (define filter-x (length (car filter)))
  (define filter-y (length filter))
  (define offset-x (/ (sub1 filter-x) 2))
  (define offset-y (/ (sub1 filter-y) 2))
  ;; compute the filtered byte array
  (for* ([x width]
         [y height])
    (define new-val
      (for*/fold ([sum 0.0])
        ([dx filter-x]
         [dy filter-y])
        (define sample-x (modulo (+ dx (- x offset-x)) width))
        (define sample-y (modulo (+ dy (- y offset-y)) height))
        (define sample-value (vector-ref data-vec (ij->offset sample-x sample-y)))
        (define weight (list-ref (list-ref filter dy) dx))
        (+ sum (* weight sample-value))))
    (vector-set! new-bytes (ij->offset x y) new-val))
  (vector->list new-bytes))

Suppose we want to rewrite that in Rust. Here’s what it might look like:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27

fn blur_rust(width: uint, height: uint, data: &[u8]) -> ~[u8] {

    let filter = [[0.011, 0.084, 0.011],
                  [0.084, 0.620, 0.084],
                  [0.011, 0.084, 0.011]];

    let mut newdata = ~[];

    for uint::range(0, height) |y| {
        for uint::range(0, width) |x| {
            let mut new_value = 0.0;
            for uint::range(0, filter.len()) |yy| {
                for uint::range(0, filter.len()) |xx| {
                    let x_sample = x - (filter.len() - 1) / 2 + xx;
                    let y_sample = y - (filter.len() - 1) / 2 + yy;
                    let sample_value = data[width * (y_sample % height) + (x_sample % width)];
                    let sample_value = sample_value as float;
                    let weight = filter[yy][xx];
                    new_value += sample_value * weight;
                }
            }
            newdata.push(new_value as u8);
        }
    }

    return newdata;
}

Pretty similar. Of course, it uses curly braces, so it runs about three times faster…

So: what kind of glue code is necessary to link the Rust code to the Racket code? Not a lot. On the Rust side, we need to create a pointer to the C data, then copy the result back into the source buffer when we’re done with the blur:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12

#[no_mangle]
pub extern fn blur(width: c_uint, height: c_uint, data: *mut u8) {
    let width = width as uint;
    let height = height as uint;

    unsafe {
        do vec::raw::mut_buf_as_slice(data, width * height) |data| {
            let out_data = blur_rust(width, height, data);
            vec::raw::copy_memory(data, out_data, width * height);
        }
    }
}

On the Racket side, it’s just a question of making an ffi call, which is super-concise:

1
2
3
4
5
6
7
8

;; link to the rust library:
(define rust-lib (ffi-lib (build-path here "libblur-68a2c114141ca-0.0")))
(define rust-blur-fun (get-ffi-obj "blur" rust-lib (_fun _uint _uint _cvector -> _void)))

(define (rust-blur width height data)
  (define cvec (list->cvector data _byte))
  (rust-blur-fun width height cvec)
  (cvector->list cvec))

And away you go!

I’ve got this code running live at FIXME. What’s that you say? You can’t seem to find FIXME?

The first Winnecowetts olympics is come and gone, and calling anything the “first” is really just asking for trouble, so I guess I’m just being optimistic.

Anyhow, it was a great success. We had 8 participants, and four events.

Target shooting

Shoot a nerf dart through a hanging bicycle tire. Three attempts at each distance.

• Gold : Christopher Moulton & Alex Clements
• Silver : Xavier Clements

Olympic Hide & Seek

• Gold : Alex Clements, Ben Taylor, Xavier Clements
• Silver : Christopher Moulton, Nathan Clements

Tower Building

Build a tower out of Kapla blocks. Each team had the same number of blocks. I think it was about 40.

• Gold : Alex Clements
• Silver : Ben Taylor & Nathan Clements
• Bronze :

Pool Noodle Javelin

Throw a pool noodle as far as possible.

• Gold : Nathan Clements
• Silver : Alex Clements
• Bronze : Alton Coolidge

Egg Roll (self)

Lying on back and grasping knees to chest, roll along from start line to finish line (about 20m).

Men

• Gold : Alton Coolidge
• Silver : Wing Taylor
• Bronze : John Clements

Women

• Gold : Liadan Taylor

As I write this, it’s Monday, August 6th. Charlotte Clews is organizing this year, and it appears that there will be two events that include the swim; the Maine Mountain Challenge, and the Granite Mon. Yesterday was the first swim. Here’s the picture:

2012 Swimmers

From left, the swimmers:

• Charlotte Clews,
• Jerome Lawther,
• Justin Pollard,
• John Clements,
• Mary Clews,
• first-time-swimmer Pat Starkey, and
• Moira McMahon

The water was extremely warm: Global Warming will kill millions of people, but it made Sunday morning easier. I think it was at or above 70 degrees. High tide wasn’t until 2:00 PM, so we started quite late. We met at the KYC at 9:30, and I don’t think we started swimming until about 11:00. We all arrived at about the same time; I got in at an hour and 28 minutes, and I think we were all in before 1:38.

There may have been a ride to Cadillac and a hike up it, but I’ll have to wait until I hear more to write about that.

{% gist 2411008 %}

{% gist 2411005 %}

{% gist 2411004 %}

{% gist 2411002 %}

{% gist 2410993 %}

{% gist 2410918 %}

{% gist 2410916 %}

{% gist 2410910 %}

{% gist 2410909 %}

{% gist 2410908 %}

{% gist 2410905 %}

{% gist 2410904 %}

{% gist 2410899 %}

{% gist 2410889 %}

{% gist 2410886 %}

{% gist 2410882 %}

{% gist 2410860 %}

{% gist 2410856 %}