Hi pypy-dev, just a bit of playing around: This afternoon I wrote a small script that randomly creates texts out of statistics collected from some input data using markov chains. It got especially funny (and quite absurd) when I took the logs of #pypy as data. Look here for an example: http://codespeak.net/~cfbolz/log.txt There are some really funny parts in there: <stakkars> I stinglike some for things like calling and how being an EU point there was on them... <arigo> makes sense. <stakkars> I said I won't have them in the results probably a question? <pedronis> if you know if we work is a dirt regression test or available <arigo> sorry for the whole lltype or: <hpk> well, profiler too, Christian to believe that something for which test? <mwh2> oh good to know <pedronis> I'm going to write a ll helper The code is here: http://codespeak.net/svn/user/cfbolz/python/random_markov.py Cheers, Carl Friedrich