These programs all implement the same basic problem. The problem is this: read stdin, tokenize into words for each word count how often it occurs output words and counts, sorted in descending order by count The idea is to measure in each language how well it performs this basic problem, how much memory it takes, and how elegant the solution looks. The test is for scripting languages, but I will gladly take non-scripting languages, too, as long as I can compile / run them on my Linux laptop on my test data to get comparable results. The data will be several hundred megabytes of web logs, so the start-up time should not matter much. I will preload the interpreter / VM for each language before the benchmark so disk I/O does not factor in, and the test data will come from a tmpfs (a ram disk, basically). The idea is to solve this problem as you would solve the problem in the language. So, for perl, for example, you'd use a hash, not implement the whole thing as a C extension to gain speed. In C++ you'd use the STL and iostreams. Since I am not an expert in every programming language, do send me enhancements if you think one implementation here is unfairly making a language look bad. I did not write all of these implementations myself, btw. Off the top of my head, I'd like to see implementations in: - Javascript (there are javascript interpreters you can run from the command line, right?) Send your submissions to felix (dash) wp (at) fefe (dot) de