At docs meeting and at docs discussions, we had a lot of discussion about documentation for Workstation Edition and there are various different positions what to do about it.
To hopefully get more clarity on this, I opened a ticket in Workstation Repository. Feel free to comment.
If we do not reach clarity in the next few days, we should make a decision at our next docs meeting (Nov. 9) according to our governance rules .
Can you elaborate on the ramifications of these?
We drop the Workstation box completely and replace it by a Silverblue box. The current “Emerging Desktops” will change to a Kinoite box. With this comes a decision not to provide any user documentation for Workstation at all. Documentation-wise, Fedora Desktop is then Silverblue and Kinoite.
We decide to provide at least a core documentation. Then, for the launch of Fedora 37 release, we need at least minimal initial documentation, such as a “Getting Started” guide. This would at least give us something for new users. I can offer to create this, provided I have at least 2 days to write it. The goal is to then complete this step by step.
The first option sounds as if Workstation were being retired entirely and Silverblue is taking its place. The second makes it sound like this is about how much to document Gnome, specifically. There are some areas that are Fedora specific, such as the Software Center defaults and enabling 3rd party repos, that might be useful for Fedora Workstation users to know about in more specificity than what the Gnome docs currently provide, for example, and this behaves differently between Workstation and Silverblue (with the latter specifically focused on flatpak).
Would it be worth documenting some basic Btrfs stuff under the Workstation edition docs (e.g. how to create a snapshot of your OS filesystem and how to restore it)? I think that is kind of unique to Fedora’s Workstation edition at the moment.
I also see a lot of hits on Fedora Magazine for information about dracut (I don’t know why).
Maybe some dracut recovery basics would be appropriate for the documentation as well. Off hand, I think recovery operations might be the most sought after thing from end users. The times I can remember people complaining about inadequate documentation seem to be when they were stuck with something that was somehow misconfigured or broken.
Be aware that there has not been a Workstation specific box before. So it would be about dropping the box from the current testing page, which means that it will not be introduced to the officially published Docs.
There is no doubt about that. There are many differences between Workstation & Silverblue, indeed. But Silverblue has already a Docs that is already in use and maintained. The problem about Workstation is: who should write the Workstation Docs? There is practically nothing yet 
So it is more about what we can do / who can do / what is possible.
Imho, absolutely 
The only page that exists so far (a rough draft without revision yet) is about that type of issue 
You can check it out here (especially in terms of btrfs, I am not experienced and it would be great if someone with experience in this area would make more out of it), but I suggest to not make changes or MR in this repo atm because I guess the branches will be shifted in the next days. Also, I generally suggest to wait for a decision of what will happen with this box/repo before putting efforts in.
Hmm, that page appears to be recommending installing third-party tools. I don’t have any personal experience with them, but I think I’ve seen a few complaints about them not working properly with Fedora Linux on ask.fp.o.
I was thinking more along the lines of some minimal command-line examples. Can’t a user just type btrfs something something ... to create a snapshot and btrfs something something ... to restore it? I thought I saw a post from Chris Murphy a while back that showed how to do that (but I’d have to hunt for it).
In fact, showing how to do the above from a dracut rescue shell might provide a good “foolproof” recovery solution in case someone totally bricks their system.
Edit: I think I found the post from Chris that I was thinking of (particularly the “if you aren’t using snapper” section):
https://discussion.fedoraproject.org/t/rollback-btrfs-root-snapshot/72074/3?u=glb
Everything is open and can be discussed I was just drafting this page to have an example of the approach to have something to discuss. There is nothing against command line in general. The automation in the end will make it necessary anyway. So there is nothing fixed. However, the backup tools are in the official Fedora repos, nothing external. btrfs, mdadm, lvm are available by default. But let’s postpone this topic to later and another topic The focus should remain for now on the Workstation guide in general.
I think, the effect of omitting Workstation again is neutral. On the current site, there is nothing about Workstation for several releases now, but Silverblue and Kinoite have each a separate box. So we would only be prolonging an already long-deficit and confusing state of affairs.
If we could at least start with workstation documentation, that would be progress and an improvement. However, that requires a commitment of the Workstation WG to deal with it (not necessarily to do the writing). Maybe the development of a documentation is superfluous, because Workstation WG is planning something completely different.
Yes, that’s a good example that we need Fedora specific Workstation documentation. And there are several Fedora Magazine articles that can help to expand the documentation. A Fedora Workstation documentation wouldn’t be a duplication of Gnome documentation.
But it depends on Workstation WG.
Is this btrfs “snapshot” even a thing? I can see taking a snapshot of a total disk where you keep files; and then taking a snapshot; and then restoring the snapshot later. But, I think people then interpret that as some sort of MacOS X time machine for the whole boot environment. Btrfs isn’t able to do that. Most Fedora installs have /boot and /EFI/boot as ext4 and fat. The Btrfs partition is kind of an LVM of /root /home /var and /temp and stuff that is meant to balloon and shrink around. Even if you snapshotted your Btrfs partition, you would never boot it without the non Btrfs partitions that don’t get snapshotted? This snapshot thing sounded interesting when first bandied about, but then anaconda never made us a snapshot-able install. So, I never really understood why the idea was ever brought up?
/boot and /boot/efi can be restored from the contents of the snapshot. But most people would need instructions on how to do that.
