Reorganizing the order of Fedora spins

This is just an idea I’m going to propose, and one that we may not end up doing if it proves to contentious.

I think we should reorganize the way that spins are presented on the website.

Below is the current order of spins on Fedora Spins | The Fedora Project

  • Fedora KDE Plasma
  • Fedora Xfce
  • Fedora Cinnamon
  • Fedora MATE-Compiz
  • Fedora i3
  • Fedora LXQt
  • Fedora LXDE
  • Fedora SoaS
  • Fedora Phosh
  • Fedora Sway
  • Fedora Budgie

As far as I can tell, spins were added in chronological order from when they were created, but I wonder if they reflect the DEs that users are interested in checking out right now. In the case of Fedora Sway and Fedora Budgie, they’re two of the four atomic desktops and yet they’re all the way at the bottom of the list and not even visible from the top bar without scrolling. You have to know that these spins exist to come across them.

Below is my proposed reordering. It’s not super scientific, but I did take a look at which are the most popular default desktop environments among a sample of distros. After Gnome and Plasma, Budgie and Xfce are the most common. I kept Sway near the top because of the Wayland support and having an atomic spin.

Sample of distros and their default DEs
Distro DE
Solus Budgie
Ultramarine Budgie
Ubuntu Budgie Budgie Modded
Linux Mint Cinnamon
Debian Gnome
Vanilla OS Gnome
Endless OS Gnome Modded
Pop!_OS Gnome Modded
Ubuntu Gnome Modded
NixOS Gnome Recommended
Kubuntu KDE Plasma
Ubuntu Studio KDE Plasma Modded
Lubuntu LXQt
Alpine Linux N/A
Arch No Default
EndeavorOS No Default
Gentoo No Default
Manjaro No Default
openSUSE No Default
uBlue No Default
elementary Pantheon
Qubes OS Xfce
Xubuntu Xfce
  • Fedora KDE Plasma
  • Fedora Budgie
  • Fedora Sway
  • Fedora Xfce
  • Fedora Cinnamon
  • Fedora MATE-Compiz
  • Fedora i3
  • Fedora LXQt
  • Fedora LXDE
  • Fedora SoaS
  • Fedora Phosh

What do you think about the idea in general? What do you think about the ordering?

1 Like

If we have some sort of official download count for each spin, I’d be OK with ordering them by that. I’m hesitant (as an individual contributor) to try to intentionally influence any ranking of the different spins. It wouldn’t be a great offense or anything. I just have a slight ideological inclination to let the spins compete on “equal” footing. I’d also be fine with ordering them alphabetically (which would incidentally help Budgie).

Edit: By the way, it looks like some of those are going away due to lack of volunteers willing to maintain them (Remove not actively maintained spins (#277) · Issues · fedora / Fedora Websites and Apps / Fedora Websites / fedora-websites-3.0 · GitLab and Phosh). So the scrolling problem might go away to some extent without any intervention.

3 Likes

The countme data includes information about variants. That’s not image downloads, but maybe better? Here’s a rough tabulation for F39 to date:

variant hits
workstation 9419722
kde 1561897
xfce 216539
cinnamon 167789
matecompiz 130753
sway 50573
budgie 40234
i3 849
lxqt 739
soas 108
lxde -
phosh -

(I didn’t see any hits for the last two)

Bonus Atomic variants:

silverblue 81814
kinoite 29772
sericea 3960

They may not be directly comparable with the others due to the different counting implementation.

3 Likes

From a Docs perspective, we definitely need to make the list neutral, i.e. without favoring one variant. That would be either alphabetical sorting or by anciennality (as is obviously the case at the moment).

In order to structure them according to possible user interest, we can also create them in 2 columns, “package based desktops” and “atomic desktops”.

2 Likes

The spins page only lists regular spins. Atomic spins have their own dedicated page (Fedora Atomic Desktops | The Fedora Project)

2 Likes

I do see that point. From a marketing standpoint my interest is more in prioritizing the spins that are more likely to appeal to new and existing users rather than holding those spins back due to history. Even from that point of view I agree that we would want something objective to peg decisions too for two reasons. First, Fedora is too big for a small group of contributors to have so much sway in the presentation of spins. Second, the maintainers of spins work hard to provide them, so treating them unfairly by simply prioritizing another spin based on preference is kind of cringe.

If some spins will be discontinued, that does help address part of the problem by having less to scroll through. Another idea I had was to copy Manjaro’s display of spins. By organizing them in a grid it makes the spins look more equal and brings more of the options higher on the screen.

Without scrolling I can see six spins to choose from:

Compare that to our page where the only spin you can see before scrolling is Fedora KDE:

@chrisawi I think the countme approach is better than what we’re doing now. Maybe that’s something that we can update after every cycle? So with Fedora 40 release we gauge the most popular spin based on this data in Fedora 39 and order that way?

I think the docs are fine and don’t need to change. The website is meant to be a showcase for what Fedora has available. Docs should not be constrained by that goal. If I had to guess I would say that many people use our docs by searching, so it’s not the same experience as a user simply going to the Fedora website to explore what spins they can use.

It might even be something we could integrate into the CI/CD system that runs (roughly) weekly.

@darknao would it be too much to add a script to the .gitlab-ci.yml that does something like:

curl -s https://data-analysis.fedoraproject.org/csv-reports/countme/totals-countme.db.gz | gunzip > totals-countme.db
sqlite3 totals-countme.db "select os_variant, sum(hits) from countme_totals where os_name='Fedora Linux' and os_version='39' group by os_variant order by sum(hits) desc;"

And use that output to reorder the contents of navigation.js? (We’d probably want to separate the downloads section out as a separate json file and import that into navigation.js instead of trying to edit navigation.js in place.)

Edit: I’m still not really sure all that would be worth the effort though. Just ordering them alphabetically would be much easier and perhaps more reliable. :slightly_smiling_face:

1 Like

Sorry to hear such news about LXQt…
My favorite spin. Small, fast, without glamour.
Installed on my laptop, my wife’s laptop, my grandson’s laptop.

ОК…
Will migrate to Budgie…

Yeah, right. I forgot that at that moment, although that was my suggestion at the time :slight_smile:

OK, I got it. So I will remove the spins from the docs landing page and leave it for labs only. For spins, there is obviously no user doc at all, anymore.

Nevertheless, Fedora is dedicated to new development and to free and reliable use of software. Fedora is not to compete for download rates or rankings. And our marketing should not chase after download figures, but rather promote and disseminate our goals and principles. And that’s why a presentation based on download figures is out of the question, either alphabetical sorting or by anciennality.

yeah, i’d be a little surprised if xfce was beaten by budgie and sway.

@joseph I think still can say that technically we can have the best from both worlds. Both immutable and mutable systems has its own advantage. But what would happen if we keeping a relatively small atomic system core that is immutable, and also a single flatpacked/docker img GUI that falls into a same category can bring a lot of advantage. A - you can kill spins and a lot of work, B - faster downloads, and setups if we separate core/userspace/customization of activities and tools. With this three level of separation, basically you can have what you want. The core level should bring you to commandline login, without GUI, the second level with gui, and the third is your own customization whatever you like. What do you think? UBlue doesn’t target similar?

If we do change to a system that is based on a performance metric, I don’t think it has to change so frequently. I don’t expect preferences to change so quickly and I don’t think the hurdle warrants as much attention as that. In my opinion, it could be as simple as reorganizing the list (if needed) once a year. I wouldn’t want to put someone out to build something complicated for this.

I wouldn’t remove Fedora Spins from the Docs landing page either. I think everything on the docs side is fine and doesn’t need adjusting based on the potential outcome of this discussion. Removing the mention of spins would be odd because they’re more popular than our labs.

I do agree that we need to keep Fedora’s values top of mind in all discussions for marketing. However, I think it’s also good if Fedora can grow in users and mindshare. By tweaking the website we’re adjusting the positioning of spins based on what is more likely to grab a user’s attention. If a visitor arrives at the spins page and sees three recognizable desktop environments followed by what look like legacy or older environments, that could be enough of a turn-off to jump somewhere else, and then they would have missed two active and important DEs at the bottom in the cases of Sway and Budgie. I don’t see it as chasing download figures. I see it as better serving current and potential users.

If the goal is to present the spins in an order relevant to a prospective downloader’s interests, then be explicit about it and gamify the process.

For example: Come up with four or five semi-quantifiable characteristics and assign corresponding coefficients to each spin. Present some sliders where users can set their own priorities for these characteristics, and the spins re-order themselves accordingly.

While I could see nerd-sniping myself out of several days work doing this, let me just say this is exceedingly unimportant. Please don’t pour anyone’s precious finite heartbeats into this unless they can find literally nothing better to do!
Fun != Worthwhile.

1 Like

I dont think anything points to LXQt being dropped? It is en par with all the other XOrg Desktops that work on a Wayland version, and I dont think Budgie is your first choice here.

I think displaying the “most downloaded last month” at the top would be interesting, could have some cool statistic chart over time too. But an alphabetical order for the rest sounds most reasonable to me.

We could also group all depending on their usage, which would help show which ones are most used and likely supported (could even show percentages) but 1. unclear if users = support and 2. unclear if this is the FOSS mentality (I mean this is done with GNOME, a lot).

Showing the atomic variant right next to the mutable one would be a good improvement I think. Its just too complex currently.

LXQt isn’t going anywhere. The Keepalive just got missed.

2 Likes

OK
Thanx

Being the maintainer of one of the spins, if this get done, I hope you put out a lot of communications explaining why and how that reason is a great reason to reorganize that.

I don’t say this in the interest of generating FUD, but knowing how this could be a demotivation to “less used”, “less clicked”. “less downloaded” and/or “worse stats” spins’ maintainers.

1 Like

There have been several posts in favor, but there was at least one against “presentation based on download figures”, so I’m still mulling over what to do. :slightly_smiling_face: I like Todd’s idea of adding controls to allow the user/reader to sort the spins by whatever criteria they prefer. But that would require some coding, so it might not happen until I have ample free time (and feel like) messing with it.

In the meanwhile, people are welcome to continue to submit ideas and votes to this thread and remind me about the issue. :slightly_smiling_face:

Considering the war between Lilliput and Blefuscu, I would leave well enough alone.

I will say, if this is an area that will make folks feel bad as @x3mboy alluded to then I’m good with dropping this. The positive might be negligible while the negative impact on the contributors could be noticeable. Not worth it I don’t think.