I
C
F
P |
ICFP Contest 2007 writeup
(13 Aug 2007 at 14:07) |
 After finishing writing my paper last week, I finally got around to writing up our experience in the 2007 ICFP Contest. It's much longer and more detailed than our previous writeups (2005, 2004, 2002, 2001), I think mostly because I'm really defensive about our totally low score. As we were last year's judges I am kinda embarrassed about this! But, well, we took a gamble on an unorthodox strategy and were bit by some last minute bugs, so there you go. Fun times, for sure. Check out the page for the full story, some images and some movies. |  |
|
Thanks for going to the trouble of writing this up!
Poor Jason! |
| Ha ha, don't worry, he always does that. |
That was awesome! Thanks!
This is one of the best (if not the best) ICFP write ups I've read (and I've read almost all of them), if only because you explained how you ended up solving the problem in a different way to everyone else :-) |
| Thanks Robert, I appreciate that! I should also mention that Spoons helped a bit with the writeup and was mostly responsible for creating the movie. But Spoons never reads blogs so he probably will never know that I neglected to mention that. |
Enjoyed your write-up. It also makes me feel so much better to know that team K.I.S.S. outscored the organisers from last year. Thanks for that! :-)
I get kind of nervous now when I hear about a new ICFP07 Contest write-up, because I'm still trying to solve this problem in my slow, plodding way - they did say it was a 72 day contest right?! This means I don't want to accidentally get any more clues, yet I also don't want to not read them!! So it's a case of skimming quickly the first time (with short term memory power turned as low as possible) and then going back to read it again more carefully if it seems "safe". Yours is nice and safe (from my perspective), and very interesting.
I wasted a huge amount time during the contest trying to track down a bug caused by an a very stupid mistake I made (and insufficient testing later). My one line "empty bucket" implementation ("bucket = []") was effectively a NOP - and unfortunately the "self check" page didn't really tell you anything useful in that case! In fact only the bottom half was visible and that (apart from the incorrect colours) seemed to show things were working well. It seems obvious now, but I ended up thinking the DNA to RNA tool was the real cause. Still, it turns out I fixed the bug accidentally when doing more testing (of the colour stuff!) and some associated refactoring very late in the contest but then threw in the towel without running the self-check again. I didn't find out it was fixed until the next morning when I started comparing my outputs to those from tools from another team.
I'm really looking forward to hearing if this year's organisers really expected teams to solve all those puzzles, etc., or whether those were really a red herring/honey pot/something to play with later when you realised time was running out and turned instead to patching the daylight image, or perhaps generating the whole thing from scratch. I guess solving the puzzles was feasible if your team was big enough (and smart enough) and perhaps also a little bit lucky, but as a one person team (and not particularly smart either!) it seems rather challenging! Unless of course, it really was meant to be 72 days... oops, look at the time... I better get back to working on a way to extract page 23! |
| Oops - just realised I called myself a one person team. Just in case my "team-mate" reads my earlier comment, we were more like a "one and a bit" person team this year. As in about 1 and 1/8th.... :-) |
| That's true, Tom, I never read blogs. |
| My head is exploding!! |
|
|
|
|