48
I asked the Linux Community to RANK 34 DISTROS, here are the results! (tilvids.com)
submitted 2 years ago by to c/linux@programming.dev
20 comments fedilink hide all child comments
you are viewing a single comment's thread
view the rest of the comments
[–] 6 points 2 years ago (4 children)

I'm surprised to see Fedora ranked so highly.

Did they provide raw scores? I'm curious about two things, one is which could be determined from vote counts, and the other which would require a different voting system:

  • I kinda want to roll up the values into the base distribution, and rank that. Distros obviously add value, but Endeavor and Cachy are a couple of extra (removable) packages descended from Arch. There's a hard, subjective version: "Is Ubuntu really just Debian with extra packages?" ; an easier version of this: "can Y be turned into base X without dramatic system modification?"; and a really easy, but probably uninteresting version: "is Y descended from X?".
  • I'm more interested in a ranked-choice version of this poll. While I suspect the results would be similar to the second evaluation version above - users are likely to rank ancestor or siblings above other base distros - it still might be interesting. E.G., I like EndeavorOS but might rank Alpine over base Arch.
  • permalink
  • fedilink
  • source
  • parent
  • hideshow 4 child comments
    •
  • [–] 2 points 2 years ago* (last edited 2 years ago) (3 children)

    I’m surprised to see Fedora ranked so highly.

    I actually think it's to be expected.

    If it was done last year, then -with the whole "Red Hat gOiNg ClOsEd SoUrCe" fiasco- it might have dropped. But most peeps seem to have forgotten about that, or just stopped caring (most prominent example for me personally would be Chris Titus; who went from an outright hater that wouldn't even touch Fedora derivatives (like the many uBlue projects) to outright praising Bazzite).

    In case one's out of the loop, these two articles by Jeff Geerling should give a complete yet nuanced take on the matter. TL;DR: Red Hat made it harder for projects like Oracle Linux to cannibalize their work, but this came at the price of closing off public access to RHEL's complete source code, and using a EULA Subscription Agreement to try to stop customers from sharing the source code. Which, actually is allowed under the GPL licence; even if some would argue goes against the spirit.

    Btw, why are you actually even surprised by it? While Fedora has historically made drastic changes that might have alienated its user base (read: being the first that went all-in on the likes of systemd, Wayland etc.), they've demonstrated to show some restraint in the last couple of years; acknowledging even that such radical changes aren't desirable. In turn, the community rewarded such efforts, making it go from "Red Hat's testbed distro" to a very respected mainstream distro. In the more recent ProtonDB reports, one can see how significantly it has managed to close the gap in usage between its ecosystem and the other big shots (read: Arch and Debian/Ubuntu).

    Did they provide raw scores?

    Yup! Here: https://nextcloud.thelinuxexp.com/index.php/s/PQPoRZo7n8dSkjw

    one is which could be determined from vote counts

    Ah, would this comment help?

    I’m more interested in a ranked-choice version of this poll.

    Me too. I suppose you could retro-actively use the raw scores for this. I'm curious of your findings!

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 3 child comments
    •
  • [–] 3 points 2 years ago* (last edited 2 years ago) (2 children)

    Btw, why are you actually even surprised by it?

    Well, because Redhat was the first linux distribution I used, and I did so four about 6 years personally and then another decade professionally (various versions, from CentOS to RHEL) and IME it's by far the worst distribution I've used, and RPM is, and always had been, a clusterfuck of a package management system. The excuse for use in Enterprise was that companies could pay for 24/7 service support, and that is often a deciding factor, especially if OPs has a strong voice in the decision process; but by god is it a horrible system.

    I'm actually pretty oblivious to any Redhat controversy; I don't bother reading anything Redhat-related anymore.

    I'm not surprised that it's widely used, for the same reason I'm not surprised Microsoft is widely used: because of the enterprise decision process. But the popularity surprises me.

    Did they provide raw scores?

    Yup! Here:

    Thanks!

    Ah, would this comment help?

    I saw that; absolute values would be preferable, but I can work with percentages - two decimal places of accuracy should be fine. It's not like we're trying to do science here.

    I’m more interested in a ranked-choice version of this poll.

    Me too. I suppose you could retro-actively use the raw scores for this. I'm curious of your findings!

    I think you can't, because it requires each voter to rank their preferences, which requires a specific form of voting mechanism. I didn't participate in the poll, but if it was run as ranked choice, and if we had access to the raw, per-voter results, and if the sample size was sufficiently large; then yeah - we could run a full Condorcet count and get some interesting answers!

    The hard part about doing an "should these two distros go into the same bucket" evaluation is determining how closely related distros are. For example, I wouldn't consider Mint to be Debian because there are no number of packages you can remove from Mint to make it pure Debian without breaking it. Believe me, I've tried. At some point, there's are very Mint-specific packages which, if you remove them, the system won't boot. A dedicated and knowledgeable enough person might be able to swap packages out and keep a running system, but the Mint-ness is woven in pretty deeply into some core package dependencies. I suspect the same is true for Ubuntu->Debian, but maybe not for Kubuntu->Ubuntu. I know you can go from Arch->Artix and back again, although it's a bit of work. I don't know if you could remove enough of EndeavourOS to get pure Arch and still have a bootable system (I haven't tried).

    So, you could just bucket everything by package manager - does it use apt? Then it's Debian! Although, now with Snap, how much is Ubuntu based on Debian anymore, anyway? Anyway, this is the last, uninteresting way.

    More interesting is bucketing by whether it's reasonably possible to convert one distribution to another. I suspect you could turn Arch into Endeavor by changing some source package lists, running an upgrade and maybe installing a package or a dozen. Figuring this out for every distribution would be hard.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •
  • [–] 0 points 2 years ago (1 child)

    Regarding what you said about Red Hat, I'm sure that -at least historically- RPM-based distros were simply a pita to work with when compared to the alternatives. I'm a relative new Linux user (only about 3 years so far), so I've only seen its better days :P .

    I think you can't, because it requires each voter to rank their preferences, which requires a specific form of voting mechanism. I didn't participate in the poll, but if it was run as ranked choice, and if we had access to the raw, per-voter results, and if the sample size was sufficiently large; then yeah - we could run a full Condorcet count and get some interesting answers!

    I'm not particularly well-versed on some of these terms. However, isn't it possible to retroactively make the gradings work as ranked choice? So, say a user gave:

    • Arch a 4,
    • EndeavourOS a 5
    • and Manjaro a 3

    Wouldn't this imply that they rank EndeavourOS higher than Arch, and both of them higher than Manjaro? Sure, we won't always have strict orderings. But I'm pretty sure this doesn't necessarily constitute a problem.

    Regarding 'distro-buckets', I think that defining a distro different from another whenever (an attempt at) applying the inverse of the changes doesn't produce a functional system is cool. I hadn't considered that before. But, as you've noted yourself, this is a gargantuan effort and (probably) not worth it. Like, e.g. let's look at Deepin, it's mostly Ubuntu with the Deepin desktop environment. However, their respective auras are very different. I think we'd lose a lot of nuance by placing them in the same bucket. Just my 2 cents*.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 1 child comment
    •
  • [–] 2 points 2 years ago*

    Yes, exactly. You need those rankings per voter - did you see that data somewhere? I saw only the summary.

    Edit Ah! The ods threw me off; I hadn't looked at it because I needed to get it on my computer.

    Yup! It looks like ranked choice, and with a little data massaging, we can get some interesting results. Stay tuned!

  • permalink
  • fedilink
  • source
  • parent