[qbasic_forever]

Go the next step and run a local kubernetes cluster with kind or k3s (it will take you 30 seconds to have a k8s cluster going). IMHO the kubectl CLI is a lot more logical than docker's CLI. You can create all your local storage volumes ahead of time, create a pod that attaches to it, and then just kubectl exec into the pod vs. writing a long fiddly docker command line string (or crafting a docker-compose.yml). It's easy to adjust the pod as necessary while it runs too, like adding a service to expose ports without rerunning the container.

But if you do like the idea of docker dev environemnts, check out a tool like batect: https://github.com/batect/batect It's somewhat like if docker-compose had make-like commands you could define. Your whole dev environment and workflow can be defined in a simple yaml config that anyone can use.

[kamac]

Won’t setting up a k8s cluster require writing resource definitions? I imagine you’d need to write a statefulset. How’s that better than writing a docker-compose? I’m not sure how vscode does it, but it allows you to publish ports in real time as well.

[qbasic_forever]

Nope you can make a simple pod definition and not worry about a deployment. For a local cluster you'll just have one node and everything is effectively a statefulset. IMHO it's easier to write k8s yaml, there's tons of tooling, clear schemas, etc. You could even script calling the API server directly.

[cyberpunk]

Check out k9s also, it’s much nicer to use than kubectl for edits and shells and things.