Round at 20:03; 27 submissions, 27 compiled, 702 games ran. Fastest game in 439 μsec, lower quartile 230 msec, median 852 msec, upper quartile 2.30 sec, mean 2.34 sec, slowest game 45.36 seconds. The graph looks pretty similar:

And a huge spike in the number of floating-point exceptions: dividing by zero isn't a great way to start a game.

So why is the analysis for this round so late? Here's a sneak peek at what I've been working on this afternoon:

All things going well, this will be the new game log viewer soon. I entirely blame Keiran for making an off-hand suggestion during the lecture today that's brought this to life. Notably, it gives better navigation between turns and rounds.

Also, fixed a bug in the player around truncating overlong player messages: one group, in particular, has very long messages, which should be fine: the player deals with that automatically, by only copying the first MESSAGE_SIZE bytes and ensuring there's a NUL at the end -- but if your message contains Unicode runes, which require multiple `char`s to encode, this will create a malformed Unicode sequence, which the JSON library we use silently drops. Hooray!

I've also added a new "game score" line to the game log. If you (like me) are parsing game logs using regular expressions and finite state machines, you'll need to account for that.

It's looking very congested at the top of the leaderboards... next round at 06:11.

UPDATE 2243: there's a bug in the game runner . more details soon, hopefully.
UPDATE 2336: there's a bug in a group who didn't read the rules. good work, team.