Soon it will take a long time to load a bot page.

Alembic

Stderr of YoloBot_(davit) was /bin/bash: java: command not found

We will need some fancy presentation slides for the kick-off, so that we can deliver a proper explaination and not appear as bunglers.

The map generator should be updated to output the format we decided upon.
It should also fill in starting positions for the players.

• Install cgroups-bin.
• Confirm that kernel has cgroups enabled. If not, try to enable them.
• Make a cgroup for bots. (One is sufficient; subgroups are possible.)
• Add an option to firejail to use this cgroups

• Explain rules
• Explain parser formats
• Login functionality
• Upload functionality

There are always two sessions running: one for the web server and one for the ranker. For some reason the session of the web server doesn't get the newly played matches that the ranker commited.

I tried fixing this by using a scoped_session and calling Session.remove() before each request. This didn't work.

It's now temporarily fixed by reloading the web server every twenty minutes with cron. (Let the shaming begin.)

We need fancy visualisations for the game!

When one of the bots does not read his stdin, the arbiter blocks.

It could be useful to have a friendly interface to control a player yourself. This way you can easily test your bot's reaction to certain types of play. A scenario editor would be great as well.
If we share code with the visualizations, this might not be too much work.

you get things like

inputparser%201.1 does not exist or does not belong to jens

or you can successfully delete them, but they are still there...

4.4G ./production

We'll run out of data soon D:

This will cause their respective pages to not be found because it will create a totally different path.

• Write blogpost
• Announce on Zeus + Cicero mailing list
• Announce on Facebook page
• Announce on Twitter
• Announce on Facebook groups

Summary: We made a visualizing ranking website. (First?) codenight on Tuesday 12th April. We will have a finale on xx May.

Right now, commands with an invalid format will crash the arbitrator. This should be fixed.

We should at least test the following

• Test with working bots
• Test with a map file with existing marches
• Test with a bot that outputs "? marches"
• Test with a bot that times out
• Test with a bot that gives incomplete output (e.g. "4 marches" and then
  only 3)
• Test with a mischievious bot
  • Marches between forts of other players
  • Marches between non-existing forts
  • Marches on non-existing roads
• Test with a bot that gives an incorrect syntax (e.g. "? mraches")
• Test with a bot immediately stops
• Test with a bot that stops unexpectedly mid-game
• Test arbiter as an executable

Things like duplicates will make for ugly error pages, they can definitely be caught and served on a friendly platter.

can we upload multiple files?
do we need to upload an archive?
what archiving software is available to unpack during 'compilation step' ?

Ideas for cool prizes? 😄

Add an option to open source your bot. That'd make it easier to brag with the sedbot.

The ELO rating system doesn't properly handle matches with more than two players. So far, we haven't organized any multiplayer matches, so it's not a big deal at the moment.

At the moment, the calculation for a change in rating for some player A in a match against {Pi} happens as follows. The mean is taken over all would-be rating changes if A battled against each of {Pi} separately.
This scheme has already been put to good use in other places, but it is made for cases where each match results in a ranking instead of -- in our case -- a single winner. E.g. a loser might have won mid-game against an opponent -- this would be hard to track -- but the rating system would account for a loss against this opponent.

We need some examples to help everyone get started.

Will @SirDestrox bot do, or will we need more?

We should make sure our implementation is correct.

How will we prevent games from going on forever? What do we do when players take to action at all, or just do stupid shit that doesn't help the game move forwards?

why is this a draw: https://zeus.ugent.be/battlebots/matches/65706 ?

Hello Zeus

I have tried to run the the arbiter with the test.json configuration, but I'm always getting syntax errors. Apparently the other members of my team don't have this issue. I've added the error output below.

$ python3.5 arbiter.py test.json

Traceback (most recent call last):
  File "../testbots/destroybot.py", line 83, in <module>
    game = Game(sys.argv[1])
  File "../testbots/destroybot.py", line 34, in __init__
    self.parse_input()
  File "../testbots/destroybot.py", line 37, in parse_input
    self.parse_forts()
  File "../testbots/destroybot.py", line 42, in parse_forts
    for i in range(int(input().split()[0])):
  File "<string>", line 1
    6 forts:
          ^
SyntaxError: invalid syntax
  File "../testbots/testrobot.py", line 144
    def orders(self) -> str:
                     ^
SyntaxError: invalid syntax

Before running the example code, I've just installed Python 3.5.1 on my linux machine (elementary OS), which you can clearly see in the output below.

$ python3.5 --version
Python 3.5.1

Is there something that I'm forgetting? (I'm quite new to Python)

Sincerely yours,
Ewout

e.g.
have a bot named
a
and a bot named
aa

you get internal server error on
https://zeus.ugent.be/battlebots/bots/jens/a

Currently there is no support for displaying the matches your bot played, with a great tool at our disposal(the visualizer), this seems a bit like a waste.

We need a nice and clear formulations of the rules for the game.

People can practically never delete a bot, since it's very actively used in the ranker. Deleting a bot causes errors in the session.commit() statement, which causes a session rollback. Bots are never deleted.

Which color represents which player?

The currently used %-win should be replaced by a more accurate ELO rating.

.strip() is called on None, crashes occur.