pouët.net

Facebook Engineering Puzzles

category: code [glöplog]
 
Has anyone taken a look at http://www.facebook.com/careers/puzzles.php?

Some of the puzzle descriptions look very easy, but in the end it's much harder than expected to get one's solution accepted, because there are runtime limits your program must fulfil - and you aren't told them, you just get a mail saying "Your program has been determined to be incorrect". Test your stress resistance...
added on the 2010-12-21 12:36:05 by Adok Adok
Sounds exactly like my experience of using / developing with Facebook then. :-)
added on the 2010-12-21 13:16:57 by gasman gasman
get a job!
I accidentally the whole facebook
added on the 2010-12-21 18:18:46 by nekomono nekomono
At first this seemed nice, but then..

Quote:
Submission directions

All submissions must execute in a *NIX type environment (sorry, no Windows specific solutions are accepted). You are not guaranteed any libraries or plugins beyond what is part of the language/interpreter itself. The following languages are accepted:


WHAT THE HELL? I sure as hell hope they accept a C99 compliant terminal/commandline application, or else risk shutting out lots of potentially interested people who code C/C++. On the other hand I might just do it on my Mac :)
added on the 2010-12-21 18:21:33 by superplek superplek
I've been taking a fast look at the problems, and they seem really similar to the problems at the IOI (International Olympiad in Informatics).

I think it was nice for the high school, but now they seem just like "blaeh, yet another travelling salesman problem" (asymmetric or not). Not interesting anymore...
added on the 2010-12-21 19:23:07 by texel texel
For example, the breathalyzer task could in theory be solved by a program that computes the Levenshtein distance for each word pair and then adds the minimal distances together. But such programs won't be accepted by the robot because they are too slow. You have to implement some sort of heuristic that weeds out words for which the Levensthein distance would probably not be minimal. I managed to speed up my program by 66 % using such a heuristic, but either it's still too slow to be accepted, or it fails computing the minimal distance for some test case (of course the test cases the robot uses are NOT published anywhere).

plek: The programs need only read from files and output to stdout.
added on the 2010-12-21 19:42:06 by Adok Adok
Well yeah I got that after 5 seconds but it comes across in such a strong manner.. Rather they had made a bit more clear, like it is: standard C/CPP application layering on CRT only.
added on the 2010-12-21 21:02:45 by superplek superplek
Quote:
WHAT THE HELL? I sure as hell hope they accept a C99 compliant terminal/commandline application, or else risk shutting out lots of potentially interested people who code C/C++. On the other hand I might just do it on my Mac :)
The keyword here is "windows specific". So for example no MFC crap. And since C99 compliant terminal apps execute in Unix, you should be fine.
added on the 2010-12-21 23:02:57 by Joghurt Joghurt
hmm plek, ga je je nou ook opgeven voor de andere puzzelclub van die cryptonazi? :)
added on the 2010-12-21 23:43:29 by havoc havoc
Joghurt doet een sterk staaltje "mosterd na de maaltijd" ;)
added on the 2010-12-22 02:05:56 by superplek superplek
http://projecteuler.net/index.php?section=problems
added on the 2010-12-22 08:23:40 by giZMo^fr giZMo^fr
I don't know if participation requires a FB account, but just because that's somewhat likely, I chose to ignore this.
added on the 2010-12-22 14:26:05 by Y0Gi Y0Gi
AFAIK you don't need a FB account, but if you have one, your results will be linked to your FB account provided you use the email address you are registered with at FB.
added on the 2010-12-22 14:32:37 by Adok Adok
havoc: Maybe the cryptonazi is in reality a cryptocommunist?
added on the 2010-12-22 22:04:12 by Adok Adok

login