OK, here's how to get Gambit on your OLPC.

As root:

yum install gcc
yum install glibc-headers
yum install make

As user olpc (the user account on the machine):

curl -O http://www.iro.umontreal.ca/~gambit/download/gambit/v4.1/source/ ( Read more... )

Hi all.

I'm an OLPC Give One Get One donor, and I got my XO-1 machine in the middle of last month. It's currently my favorite toy -- it exceeds even the Wii on the scale of "cool geek things to have".

Today I managed to compile and run Gambit Scheme on the thing, and I also managed to get into the Smalltalk guts of the included Squeak eToys ( Read more... )

Currently all there is of it is a semi-tested gap buffer implementation. But you know how Microsoft was saying they're going to build an Emacs.NET? Well, so am I ( Read more... )

I grabbed some free fonts for Lunix that look like popular commercial fonts (like Optima). Problem is, they look fine big but if you shrink 'em small they look like ass due to crappy hinting ( Read more... )

I have removed the SpriteElement semantics from SpriteCore. Sprite inherits directly from SpriteNode. Separation of concerns is still maintained with a new kind of object called a SpriteAgent. These are pluggable behaviors that can change how the sprite moves and is drawn. Sprite delegates its behavior to one agent; and its appearance to another. ( Read more... )

Hey, remember that CPS converter I wrote a while back? Well, it sucked.

Granted, I have an erection for the CPS transform itself. By making continuations explicit we not only tail-call optimization easy, we also enable the fast implementation of call-with-current-continuation and other control constructs like delimited continuations ( Read more... )

Okay, digdoggers, this should be my final revision of the Gambit dataflow implementation. And I think I have it right this time.

It turns out that Gambit does indeed have an unbound-variable datum: it's written simply #!unbound. This is totally not documented in Gambit's manual, so it may disappear in a future release. The thing is though you can' ( Read more... )

Sorry for the dearth of code, recently. I was trying to make this a developer's blog unlike a great many other developer's blogs, wherein I showcase actual code I've written instead of simply musing about whether or not FizzBuzz is a legitimate interview problem ( Read more... )

More crap. This is a version of the dataflow program from the previous entry that uses Gambit's unspecified value (written as #!void) to indicate that the dataflow object is unbound. I'm not yet sure which style is considered more correct, but I think I like the ability to declare my own type better, the reason being that #!void is intended to be ( Read more... )

Here's some more crap. A simplistic, stupid implementation of dataflow variables (think Oz) for Gambit. I should call them dataflow cells or dataflow objects really; as currently they have nothing really to do with variables. I have some ideas for a hypothetical metacircular evaluator/compiler in which you can declare but not bind variables; such ( Read more... )