Goodbye SnapCI (and other news)

So after my gushing post about the wonders of Snap, ThoughtWorks shut it down. The official, off the record line was that they didn’t have the user engagement numbers to justify the development efforts. I can also understand that they wanted to simplify the portfolio before they were sold to private equity.

Moving on from Snap was a little jarring. No other tool that I could find implemented the pipeline the way I expected to. CircleCI 2.0 worked out OK. I found the workflows confusing at first: you need to copy artifacts around rather than lazily expect them to be there. Config is in your project in a YAML file, so I was able to model the workflow in a simple Make-like structure:

workflows:
  version: 2
  ff-ami-staging-prod:
    jobs:
      - fastfeedback
      - ami:
         requires:
           - fastfeedback
      - staging:
         requires:
           - ami
      - approval:
          type: approval
          requires:
            - staging
      - production:
          requires:
        - approval

I didn’t need Circle’s Docker support for this project. Managing dependencies in a container is a justifiable use of Docker in my view. Faster than the previous approach of spinning up AWS instances, and a world away from configuring those dependencies by hand.

In other news

CircleCI has raised 31 million bucks last month, so I think you can expect them to be around for a while.

CloudBees have announced that they’ve acquired CodeShip. CloudBees have raised 51 million in total. Amazing when you consider the very humble roots of CI.

I should be working on an investment pitch for a CI service aimed at VB.net programmers in the landscape gardening industry. TreeCI anyone?