Preferred engine per instance, where to store setting

[afbjorklund]

Should we add a value to hold the default container engine, for each instance ?

e.g.
default: "containerd" (nerdctl)
docker: "docker"
podman: "podman"
ubuntu: "containerd"
alpine: "" (none)
fedora: "containerd"
etc

Or what would be the best way to "know" which client to use, for each instance ?

nerdctl.lima
docker.lima
podman.lima

---

The main use was for the Lima GUI, where it should be possible to interact with it.

• Add basic interaction with containers and images afbjorklund/lima-gui#18

Examples:

• Version
• Info
• Containers
• Images

But also for the CLI, it could be useful with a single wrapper for all three engines.

Basically: limactl shell $INSTANCE $ENGINE

[jandubois]

I don't understand how it would be used. Is this a data that limactl ls --format would expose? In that case I think we need to re-open the limactl inspect discussion. 😄

Anyways, if that is the idea, then I think it would be better to have a more generic mechanism, like

metadata:
  engine: containerd
  cli: nerdctl.lima
  ...

So you can define whatever additional fields you want. Of course you still need some kind of convention, so that more than one image exposes the same metadata field; otherwise it becomes pretty useless.

[afbjorklund]

I'm not sure the best way to get to it, but I guess so - reading the JSON but anyway...

I could input it with a drop-down in the GUI, or "try everything" like is done with CRI.
But I think the best would be if the instance store already contained the information ?

Currently I have just added data sources like uname, uptime and /etc/os-release.

---

Currently these are used: limactl list --json

nerdctl version --format=json
nerdctl info --format=json
nerdctl container list --format=json
nerdctl image list --format=json

Was thinking that these could be generalized ?

It isn't ideal, sometimes the syntax is '{{json .}}'

$ docker version --format json
json