Blog-O! (Posts about scheme)

Some notes on cross-compiling GambitC

Blake Winton — Fri, 12 Dec 2008 22:09:47 GMT

The command to use is:

$ env CC=/usr/local/bin/arm-apple-darwin-gcc CC_FOR_BUILD=gcc ./configure --host=mac; make

Well, kind of. First you do that, then you copy gsc/gsc to gsc/gsc.onboard, then you go to a new directory, and type:

$ ./configure;make

and copy the gsc/gsc from that directory into the first directory.

To compile a script into an exe:

$ gsc/gsc -:=. -c euler.scm
$ gsc/gsc -:=. -link euler.c
$ /usr/local/bin/arm-apple-darwin-gcc euler.c euler_.c -Iinclude -Llib -lgambc -o euler

It's freaking huge!

$ ls -alh euler
-rwxr-xr-x   1 bwinton  bwinton  4M Jan 23 14:22 euler
$ ls -alh /WifiToggle
-rwxr-xr-x   1 bwinton  bwinton  17K Jan 16 14:07 /WifiToggle

And it's not a lot faster. 0.1643 seconds for the compiled version, as opposed to 0.1803 seconds for the interpreter.

But on my Mac:

$ more m1.c
power_of_2 (int x) { return 1&lt;&lt;x; }
$ more m2.scm
(c-declare "extern int power_of_2 ();")
(define pow2 (c-lambda (int) int "power_of_2"))
(define (twice x) (cons x x))
$ more m3.scm
(write (map twice (map pow2 '(1 2 3 4)))) (newline)
$ gsc/gsc -:=. -link -flat -o foo.o1.c m2 m3
$ /usr/local/bin/arm-apple-darwin-gcc -Iinclude -bundle -D___DYNAMIC m1.c m2.c m3.c foo.o1.c -o foo.o1
$ ls -alh foo.o1
-rwxr-xr-x   1 bwinton  bwinton    13K Jan 23 14:45 foo.o1

then on the iTouch,

# scp bwinton@latte.ca:/Users/bwinton/Programming/Bazaar/iTouch/gambc-v4_1_2/foo.o1 .
# gsi foo.o1
((2 . 2) (4 . 4) (8 . 8) (16 . 16))
# gsi
Gambit v4.1.2

&gt; (load "foo")
((2 . 2) (4 . 4) (8 . 8) (16 . 16))
"/private/var/root/foo.o1"
&gt; (twice 5)
(5 . 5)
&gt; (pow2 10)
1024

An odd restriction

Blake Winton — Wed, 09 Apr 2008 14:20:43 GMT

As I mentioned before, I’ve ported Gambit Scheme to my iTouch, and have been playing around with it a little. It’s pretty nice all in all, but I recently ran into a small problem while I was trying to play around with macros. The problem? There’s no way to enter a backtick (`) on the iTouch! That means that I don’t really have a way to write code like

1	`( a b ,(+ 1 2) d)

which makes writing macros a lot more painful. Fortunately, I got a lot of help from the people on IRC and on the Gambit mailing list. Specifically, Marc Feeley, the author of Gambit, posted a snippet of code that I could put into my .gambcini file that would add $ as a synonym for `. The code looked like this:

(begin
  (##readtable-char-class-set!
    (current-readtable)
    #\$  ;; the character to dispatch on
    #t   ;; this character is a delimiter
    (lambda (re c) (##read-quotation re #\`)))  ;; handler
  #f)

and the example, which works, is:

1 2	$(1 ,(+ 2 3) 4) output: (1 5 4)

If you’re trying to write that code on your iTouch, you might notice that it includes the forbidden `, and so you’re once again out of luck. Except that in this case, you can replace #\` with #\u0060, which you can type in on the iTouch, and then it’ll all work.

More stuff on the iTouch.

Blake Winton — Sat, 19 Jan 2008 02:06:51 GMT

Yeah, like I was done messing with my iTouch. Whatever.

I've been playing around a lot with the iTouch, and developing quite a few programs for it. The first one was by request, although I’m also finding it very useful. (And apparently a few other people are as well! Cool!) It’s called WifiToggle, and it’s a small app that simply toggles on and off your Wifi connection, so as to save batteries. The next thing I did was a port of the SpiderMonkey Javascript interpreter. It works, and it wasn’t that hard to get compiled and installed, which is really pretty cool when you stop to think about it.

The final thing I got working just tonight (which I was playing around with mostly to take my mind off of stuff. See tomorrow’s post for details), was a port of the Gambit Scheme interpreter. I know, Scheme is just like Javascript, only with an uglier syntax, but I still like it, and hopefully the lack of punctuation will make it easier to use on the iTouch. Heh.

Anyways, if you own an iPhone or iTouch, and have jailbroken it, be sure to add http://iphone.latte.ca/ to your list of Installer sources, and geek out!

Isn't that just the way of it...

Blake Winton — Wed, 09 May 2007 21:29:00 GMT

I mentioned in the previous entry that I was missing Python's List Comprehensions. Well, I've gotten a little closer to having them. For some reason, LispMe doesn't come with a zip method, and you can only get map by importing a "Standard Library" memo. So, I had to write my own version of zip, and here it is, for anyone else who might find it useful.

(define (zip s1 s2)
  (if (or (null? s1) (null? s2)) '()
    (cons (list (car s1) (car s2))
      (z (cdr s1) (cdr s2)))))

Along with my halfway function, redefined to be (define (halfway x) (/ (+ (car x) (cadr x)) 2)), I can now write (map halfway (zip current-point next-point)) to get the point halfway between where I am, and where I am going to.

I also came to another realization. I was planning on defining a current-point, and using set! to update it to the new halfway point, but when I think about it, I don't really care what the current point is at any time other than processing, so there's no particular need to set it, and instead I should just pass it around as a parameter, making my iter function (and I just made up the name "iter function", based on the <a href= "http://mitpress.mit.edu/sicp/full-text/book/book-Z-H-10.html#%_sec_1.1.7"

sqrt-iter function in "Structure and Interpretation of Computer Programs") look something like this:

(define (triangle-iter start num-iters)
  (if (positive? num-iters)
    (begin (draw-point start)
    (triangle-iter
      (map halfway (zip current-point next-point))
      (- num-iters 1)))))

Does that look appropriately Scheme-ish, do you think?

Starting Scheme.

Blake Winton — Wed, 09 May 2007 21:28:58 GMT

(Posted to the LispMe@YahooGroups.com list) Hello.

Let me start with a brief introduction. My name is Blake Winton, and I’ve owned a PalmPilot since back when they were called PalmPilots and made by USRobotics. I’ve been programming for them for years, mostly in C and C++, with brief excursions into Lua, Python, and Forth. My day job consists mainly of Java (J2EE), with some Python when I get the chance, and some JavaScript when I shouldn’t be using Python. On my off hours, I read a lot about other languages (recently Objective C, OCaml, and Ruby). Now I want to learn Scheme, and thought that LispMe would be a good way to do it. I’m finding some things are tripping me up, mainly due to my attempts to transfer my knowledge from other areas into LispMe. (One piece I miss in particular are something like Python’s List Comprehensions, which I believe they sole from Haskell.)

So, I’m going through SICP, but while I’m doing that, I thought I would try to do a "Real World" (tm) task, and write a program that drew the Serpinsky Triangle, using an iterative, random, approach, (Details available upon request,) and I’ve run into some small questions about best practices, or basically how to do some simple things. In return for this help, I’ll create a sort of Tutorial document for LispMe that will be able to help other people get up to speed. (If such a document already exists, please, someone, point me to it!)

So, here’s what I’ve got so far. It doesn’t work at all, but it’s starting to take shape.

; Triangle
(define points
  #((80,10) (10,150) (150,150)))

(define (next-point p) (
  (vector-ref points
    (random (vector-length points)))
  ))

(define current-point '(80 80) )
(define (halfway a b) (/ (+ a b) 2))

And we’re done. Any comments on it, from spacing to indentation to whether I should use a vector or a list for the points themselves, would be greatly appreciated. If it matters, I plan on extending the point to include red, green, and blue data as well, possibly with accessors, looking something like this:

(define current-point '(80 80 255 0 0))
(define (x pt) (car pt))
(define (y pt) (cadr pt))
(define (r pt) (caddr pt))
(define (g pt) (cadddr pt))
(define (b pt) (caddddr pt))

Oh, and I suspect I’ll be blogging my progress at http://weblog.latte.ca/blake/tech/scheme/ (Nothing exists there at the moment, though. Give it time.)

Thanks,
Blake.