The deadline for GSoC applications is fast approaching. I’d totally forgotten about it until the obligatory CoC thread, so I hadn’t given it the slightest bit of thought until the last couple of days.

Now that I remember it exists though, I really want to enter. Despite my recent, um, problems with anything resembling writing a program, and the fact that to date I have not written a single program exceeding about five hundred lines. I have to start somewhere, I suppose? A GUI for darcs, written in Haskell, is probably not the best place, though. But it would be fun!

Pros

• Something to do
• Might actually *gasp* do something useful
• Helping darcs
• CV++
• Money :v

Cons

• Difficult
• Requires attention span
• Takes time
• Hindering darcs with half-arsed contributions

So, will I apply? Ideally I’d say I’ll consider it. Really though, inertia means I probably won’t. Maybe next year.

edit: Obviously I don’t absolutely positively have to do a darcs GUI. It’s just something that caught my eye. Also Boost.SafeInt, which looks pretty easy, but if it’s as easy as it looks to me then someone with actual C++ knowledge would’ve done it in about an afternoon, so it’s probably not.

Factor is a fun language. It’s postfix, making your code look like mirror-image LISP (or, RPN, which of course is what it is), and for the most part, all the data (and functions, etc) are all on a single stack at once.

An example, which finds the answer to Project Euler problem 5:

USING: math kernel math.ranges sequences ;
 
: gcd ( x y -- z ) dup 0 = [ drop ] [ swap over mod gcd ] if ;
 
: lcm ( x y -- z ) 2dup gcd / * ;
 
: answer ( -- n ) 2 20 [a,b] 1 [ lcm ] reduce ;

Take the gcd function. It looks at the top of the stack, and if it’s equal to zero, deletes it. We have to duplicate it before checking, because = (and < and friends) replace the top two elements with t/f, and we may want to keep it around. If it’s not equal to zero, we swap the top elements, copy the second one over to the top, find the modulus of the top two, and recurse. So it’s basically the Euclidean algorithm in RPN.

Isn’t it funny-looking? The spaces around the brackets and (semi)colons are necessary; the tokens are separated by whitespace but otherwise can contain any character. Like in the above, [a,b] is all one word, a function which makes a range (or interval if you use the one in math.intervals, of course) from the top two stack elements.

For bonus points, it supports functional programming; the blobs in square brackets are ‘quotations’, which are just pushed on the stack as they are to maybe be called later. We need to do this for if, because only one of them is ever used. Similarly for while the quotations (one for the condition, one for the body) are called repeatedly.

A pity my syntax highlighting doesn’t support it.

edit: Euclidean algorithm fixed.