The Problem

You push (or better, merge) to master, and the deployment happens automagically. No further work required. That's the promise of GitHub-Actions (GHA), and probably every other CI service provider.

However, things are not as shiny as they seem at first glance. There are many problems with the approach that GHA and co. use. Some are addressable by more disciplined usage behavior¹, but others are not.

My first and foremost problem with them is: you cannot run them locally. There are some projects, that attempt to enable you to do that, e.g. act, but they are very limited. And that is because of my second-biggest issue with GHA: the runner is closed source.

Hosting a git repo is hardly more than providing a file system and SSH access. The actual mechanism they use to keep you on their platform is the CI-pipelines (and maybe the issue system and wiki, but less so). Half a year ago, I started porting my teams Azure-Devops pipelines to GHA, investing ~1 day per week. AND I AM STILL NOT DONE.

The fact that you have to push a commit to the repo to test a change, and then wait for a runner delays everything indefinitely, and makes iterating really painful.

Additionally, writing pipelines in YAML is just painful. It's super verbose, and you have a lot of "code" duplication. Quite soon in the process I invested 3 days to write a small DSL that compiles to GHA². This already took care of the code duplication, and I integrated actionlint to reduce the number of pushes I'd need. Sadly, actionlint catches a similar percentage of your bugs like a C-compiler (which is not very high). Originally, that DSL hat two backends: One to compile to GHA, and one to run stuff locally. The second one is not in use anymore. Why? Because there are tons of actions³ that you want to use, e.g. for terraform. To use those locally, I'd have to reimplement the complete GHA runner, or at least, locally reimplement all the actions I want to use.

Many of those actions are written by 3rd parties, not by GitHub/Microsoft, but it benefits them a lot by strengthening their position against their competition. It's a network effect. And it's bad for everyone except MS. I'm only somewhat knowledgeable in Azure-Devops and GitHub actions (much to my own dismay), but I assume that more or less every CI provider does its own thing, and there is a lot of duplicated efforts.

An attempt at a solution

The base idea is the same for every CI provider: You define a "workflow" that runs some scripts. Some of which run in parallel, some of which require others to finish first. They need access to a git repository, and they might have to interact with it. Defining a language to run scripts in parallel with some constraints is actually quite easy if you just pass the script to an existing interpreter and let the language mainly deal with the constraints around order and parallelism (and code reuse). However, if this is then just a new solution, little is gained. So it would be necessary to have different "generators" that generate small wrappers for your CI-provider of choice that call the actual definitions, and maybe to call a subset of functionality that is shared by more or less everyone (e.g. creating issues, merging a branch, etc.). If people who currently write "plugins" for different CI providers instead focus on this hypothetical language, we would end up with an ecosystem of CI tools that can run everywhere, including your PC. I'm thinking of something like this:

And yes, I'm aware of XKCD #927, but It's actually not the same as a CI-Provider, and I'm not aware of a similar project. Corrections are welcome. And while I'd love to work on this, I don't see myself having enough free time in the foreseeable future to do this. So I hope, someone feels inspired.