Slacker Cow 🐮

Hubot instance for Khan Academy, Slack Edition.

Differences from Culture Cow

Culture Cow communicated via XMPP for Hubot, but used a custom HTTP adapter to send formatted messages (fancyMessage) back separately via the HipChat API.

The Slack Hubot adapter uses their realtime API, and should support formatting.

To make nicely formatted Slack messages, see the following:

• https://api.slack.com/docs/formatting
• https://api.slack.com/docs/attachments

Discussion of how to send attachments via hubot-slack:

• slackapi/hubot-slack#170

Also, several things that were formerly in culture-cow are in separate repositories. For bovine culturey goodness, head over to Khan/culture-cron. For simian deploy magic, head over to Khan/khan-sun.

Making a deploy

Prerequisites

You will need a working gcloud tool and a Docker environment for deploying.

• Set up the gcloud tool as per instructions. (Note that on Mac you can use brew cask install google-cloud-sdk instead of their installer, if you prefer.)
• If you wish to use the shell adapter, get a working Docker environment setup:
  • On a Mac, the absolute easiest way to do this is via Docker Toolbox, also available via brew cask install dockertoolbox, and follow their "Installation Guide" guide if you are new.
  • For Linux, find instructions for your distro.

Development

You can run hubot as a shell instance locally, without requiring an adapter.

To make things easy, a Docker Compose configuration is provided for running things locally. Simply do docker-compose run hubot and slacker-cow will be fired up in a local shell (compose will make sure a redis image is downloaded, activated, and linked to the hubot container if required).

Once loaded, you can interact with it as if you were in a Slack session:

$ docker-compose run --rm hubot
slacker-cow> [Mon Aug 31 2015 15:47:34 GMT+0000 (UTC)] INFO hubot-redis-brain: Discovered redis from REDIS_URL environment variable

slacker-cow> slacker-cow: ping
slacker-cow> PONG

The docker compose configuration automatically mounts the /hubot/bin and /hubot/scripts directories in the container as volumes, so that you can test code modifications without rebuilding the image.

Deploying

• Run deploy.py.