The PR https://github.com/backus/rubocop-rspec/pull/205 … was reviewed by @lukmartinelli (Docker linter author) + @dkubb (one of the most thorough engineers I've met)
-
-
-
I knew very little about docker before so Hadolint (http://github.com/lukasmartinelli/hadolint …) and Shellcheck (http://github.com/koalaman/shellcheck …) helped a lot!
-
I mention the code review because I think it's a great resource now. I documented what I could and
@dkubb +@lukmartinelli taught me a lot! -
The setup uses
@buildkite's Elastic CI (http://github.com/buildkite/elastic-ci-stack-for-aws …) and a docker ruby matrix (http://github.com/backus/buildkite-ruby-matrix …). Costs me about $10/mo. -
I keep a 2nd CI up for 3rd party PRs. Buildkite still runs but they dont have public build pages yet. Most PRs are from collaborators anyway
-
I hard code a build step to check only I modify build config. Code after that runs inside an isolated Docker container with a timeout.pic.twitter.com/qX84OGuifn
-
Not saying it is totally secure but I think it is good enough. If people start sending spam by opening malicious OSS PRs I'll reconsider.
End of conversation
New conversation -
-
-
I meant to say 3 min 15 seconds for travis, not 5 minutes. Whoops!
-
I've seen even more drastic differences between "generic hosted" CIs and "custom bring your own computation resources one".
-
The fundamental difference is that the "provided resources CIs" simply suffer from bloated base images. Adds tons of latency.
-
yup. Not sure if it is Travis, Circle, or both, but every day I would watch them download and install ruby from source.
-
makes sense for a CI that can run any project instantly, but
@buildkite makes it easy to spend a day and cutout all that waste
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.