Remove usage of rpm-ostree

[tulilirockz]

As you all know, rpm-ostree is being "deprecated" in favor of bootc and Bluefin should follow that lead, this is just a tracking issue for what I think needs to be done so that we completely do not depend on rpm-ostree anymore

• [ ] rpm-ostree cliwrap being used in main
• [ ] rpm-ostree being used in main
• [ ] rpm-ostree calls in the build scripts (e.g.: https://github.com/ublue-os/bluefin/blob/main/build_files/dx/03-packages-dx.sh) (https://github.com/ublue-os/bluefin/pull/1954)
• [ ] rpm-ostree calls in the user-facing ujust & /usr/libexec scripts (bluefin-system.just and bluefin-apps)

After removing rpm-ostree/moving to fedora-bootc

• [ ] Add bootc container lint to the final build steps

[RealVishy]

Would this be as simple as replacing rpm-ostree with dnf in some of these examples?

[Malix-Labs]

dnf is not an alias to dnf5 yet

dnf has bindings to rpm-ostree layering mechanisms (not dnf5 yet)

Maybe that's something to do in upstream ?

Otherwise, probably to wait for upstream to make dnf5 reach feature-parity with rpm-ostree There are open issues about it in upstrem issue tracker too

[castrojo]

Would this be as simple as replacing rpm-ostree with dnf in some of these examples?

Yes, bsherman is testing in his custom image: https://github.com/bsherman/bos/tree/main/build_files

We need to methodically replace rpm-ostree calls with dnf5, same with the copr calls. We should probably do it a little bit at a time, we need to ensure that replaced calls pull in the same packages as we're not sure if there's a behavioral difference.

[RealVishy]

I'm happy to work on the user-facing scripts, once I've got a local registry setup to test on!