leeroybot config to get things started

gistfile1.txt

[11:00:20] $ heroku config
=== leeroy-bot Config Vars
HUBOT_IRC_DEBUG:             1
HUBOT_IRC_NICK:              leeroybot
HUBOT_IRC_NICKSERV_PASSWOR:  ---------
HUBOT_IRC_NICKSERV_USERNAME: leeroybot
HUBOT_IRC_PORT:              6697
HUBOT_IRC_ROOMS:             #dangerzone,#breakpad,#airmozilla-dev
HUBOT_IRC_SERVER:            irc.mozilla.org
HUBOT_IRC_UNFLOOD:           true
HUBOT_IRC_USESSL:            1
HUBOT_JENKINS_AUTH:          bots:-------
HUBOT_JENKINS_URL:           https://ci.mozilla.org
HUBOT_TRAVIS_DEBUG:          1
REDISCLOUD_URL:              redis://rediscloud:----------@pub-redis-19450.us-east-1-1.2.ec2.garantiadata.com:19450

gistfile2.md

github change --> travis build --> notification to heroku hosted robot.

so

your github project has a .travis.yml with a notifications section. see: https://github.com/mozilla/socorro/blob/master/.travis.yml#L36

your hubot fork has a .travis.yml with a heroku deploy section. You need to set that up with the travis-ci client library and your heroku info. see: http://docs.travis-ci.com/user/deployment/heroku/ , https://github.com/lonnen/leeroybot/blob/master/.travis.yml#L4

you should change the irc user. the readme has lots of info about how to write a custom script. When I need something slightly wonky I tend to steal the script from the main travis scripts repo (linked in the readme) and add a couple lines of IF/ELSE/THEN/UNLESS logic to filter out noisy I don't need.

This set up will auto deploy on PR merge. If no one interacts with the bot for a few hours heroku will spin down the single node and the bot will log off. You can work around this by throwing a monitor on, or curling with a cron, the help URL for the bot. http://your.domain.com/:botname/help

For writing coffeescript:

http://coffeescript.org/

I heavily use the "try coffeescript" section at the top of the page. It will live transpile, show you errors, etc.