"properties": {
        "live_rootfs_fstype": {
          "descripton": "Filesystem of live-rootfs.img",
          "type": "string",
          "enum": [
            "erofs",
            "squashfs"
          ],
          "default": "squashfs"
        },
        "live_rootfs_fsoptions": {
          "descripton": "Parameters for generating live-rootfs.img",
          "type": "string",
          "default": "-comp zstd"
        },

We discussed this in a meeting last week with @mvo5. What we (CoreOS team) would really like to do here is actually define these settings inside of the payload that is the CoreOS (Fedora CoreOS or RHEL CoreOS) filesystem/target image.

The question we have here is: can we inspect and do things based on the target image?

If yes then what's preferred? Should we base our behavior on:

• target OS versions?
  • Seems hacky to have business-y logic as part of an osbuild stage
• target OS capabilities?
  • i.e. if erofs is supported then do erofs
• some configuration files inside the target OS?
  • (e.g. bootc print-config/or file)

Our preference at this point is to base it on some configuration intentionally applied inside the target image.

We discussed this in a meeting last week with @mvo5. What we (CoreOS team) would really like to do here is actually define these settings inside of the payload that is the CoreOS (Fedora CoreOS or RHEL CoreOS) filesystem/target image.

Right, i've opened this PR to start the discussion. Reading configuration file(s) is in progress right now.

@mvo5, @dustymabe , @jlebon as discussed, 2nd patch here adds ability to get options from file

I'll review and test this today. Thanks @nikita-dubrovskii