There's Snap CI, which is by ThoughtWorks (just as Go CD), and that's in the cloud. I don't know if they use the same code base, but it might be worth investigating if they can deal with multiple repos in a sensible way.
- Plain jobs (disconnected dots) (Jenkins/Travis-CI/etc)
- Full-on graphs with value stream maps and arbitrary connections (GoCD/Concourse/etc)
- Isolated pipelines (disconnected matrices/lists of columns) (Snap-CI/GitLab CI/etc)
I've lamented for a long time that there was no "lightweight" self-hosted pipeline tool that makes full use of modern tech (lightweight API, separate UI, offload the actual build environment to containers) and does nothing more than linking the pipeline and gets out of the way. Most organizations just don't have the engineering skills to handle build graphs. And the alternative job-style ends up in a complete mess.
Screwdriver fills a real void. I'm not a fanboy by any chance, but I'm really thankful to Yahoo for opening up this void so that many will follow.
Btw, Jenkins 2 Pipelines are not even in the picture because are neither proper pipelines nor proper jobs. It has no resumable/pausable stages, and is just a glorified DSL/UI on top of a multi-job kind of model.
That glorified DSL is Apache Groovy. Its backers have been trying to distribute it by getting other JVM-based products to bundle it as a DSL, without worrying about how closely it really matches the product's functionality, such as specifying pause/resume in pipelines.
I've never missed simple shell scripts, builds triggered by commits, tailing logs so much. Being closer to the metal. Back then, I had no idea the obvious solution was called "continuous integration".
GoCD has a pretty dash board though. So I guess that's progress.
No it's not, because scripts are better to debug the problem when (invariably) all goes to shit.