What Linux Distribution Should Be Benchmarked The Most?

[mrugiero]

This is just false, Wayland is in the default Ubuntu repos.

Being supported and being on the default repos are two different things. Anyway, as for benchmarking, the only thing we need is the thing working correctly (where correctly == not worse than in any other distro), not having extensive QA. The thing is, you will need to test a desktop other than Unity, and for me Ubuntu (as in flavor) == Unity.

[mrugiero]

Fedora and Arch Linux. Ubuntu with his pro-Mir patches that won't be included anywhere else is a distortion to any relevant benchmark of the Linux graphic stack.

"Relevant" is very relative. If you mean relevant for judging the general Linux graphics stack, then yes, you are right. If you mean relevant for general public, for most of them Linux is not relevant at all, and for the ones it is, for the most (I'm talking about regular Joes that want to switch from Windows or recently did, here) whatever Ubuntu use is far more relevant than the general Linux stack, since that's what they are likely to use. That's why I think both should be benchmarked.

[GreatEmerald]

LXDE might not stick strictly with X, but Lubuntu maintainers want to avoid the extra memory use compositing implies (from the buffers you need for every window and such, not implying that either Mir or Wayland would use more memory by themselves because that's not even likely). Remember one of the focus of Lubuntu is to minimize memory use, and they already stated they might want to avoid compositing all in all. Wayland and Mir mandate compositing for traditional desktop use.
Also, Openbox might not be ported (it's not very active, so I don't think anyone will port it in the short term), so they'll have to use a different window manager than the de facto standard for LXDE if they want to use Wayland.

In the short term, yes. In the long term it may be too difficult for Lubuntu maintainers to keep both standalone X.org and LXDE on X.org running correctly.

Yes, all the Ubuntu derivatives could rebase on something else, but how could they do that seamlessly so that users can upgrade from their existing installs? It certainly isn't trivial to write an upgrade program that takes arbitrary Ubuntu derivative desktop install X and replaces it with Fedora whilst maintaining all the user's custom configuration and packages. I suppose they could take the easy way out and not allow users to upgrade - demand a fresh install - but that's not user friendly. There is also the issue of converting all of their custom package patches, package/ISO build scripts, installers, etc. to a new distribution. It could be done, Debian would be the obvious choice. For some derivatives the switch might be trivial, but for some others it could be quite a bit of work.

But why should the people who make those Ubuntu derivatives bother switching? I doubt their users are demanding it. Perhaps they are still happy using Ubuntu. Should they do it to satisfy the whims of Phoronix users? Keep in mind that Xorg isn't being removed on Ubuntu, so any reason you propose has to be better than the default of carrying on using Ubuntu+Xorg. The only possible reason I can see is if Ubuntu ceased as a distribution - but that is unlikely - it is much more likely to morph into something else than to disappear altogether from the face of the earth.

You're missing your original point. You claimed that there are a lot of distributions that base off Ubuntu, and thus Ubuntu is needed by the rest of the GNU/Linux platform. And the answer to that was that those distributions could have been based off Debian just as well. You're talking about something else here.

This is just false, Wayland is in the default Ubuntu repos.

Wayland is in the default Debian repos. Ubuntu just relays the package.

[LinuxGamer]

As for 12/2011

Facebook Shares
Another interesting statistic we can find online is the number of times a distro?s home page has been shared on Facebook. While clearly not an exact indication of a distro?s popularity, it can give us some idea of how many people share a link to their favorite distro with their friends. The following statistics come from the Facebook Graph API and show links to the distro?s homepage. Ubuntu has a clear lead in Facebook shares, with Linux Mint and Fedora fighting for second.

Ubuntu - 83,945 shares
Linux Mint - 7,762 shares
Fedora - 6,313 shares
Debian - 3,986 shares
Arch - 1,445 shares
CentOS - 979 shares
openSUSE - 599 shares
PCLinuxOS - 573 shares
Puppy - 426 shares
Mandriva - 419 shares

This is like a fanboysm meter.

http://www.starryhope.com/ubuntu-mos...-linux-distro/

But i think this is more accurate:

http://linuxcounter.net/distributions/stats.html

Ubuntu 28,5%
Debian 19,5%
Fedora 7,4%
Slackwave 6,4%
Suse 6,2%
Gentoo 4,2%
CentOS 4,1%
Arch 3,5%
Kubuntu 2%
Red Hat 1,7%
Mint 1,6%
etc.

thats for registered users...
"Of course, this way you won't get all Linux users counted as not every Linux user will register himself at the Linux Counter site."
the Facebook Graph's hmm, Ubuntu has facebook preinstalled and, i know damn will it's not counting Tv's Android etc that run Linux,
i see you don't want to use the Real Sites that show Ubuntu being at sub 1% of Linux Users, i say Ubuntu is more like 20%-30%, i know even on the sites i have admin to it shows Ubuntu being really low on the counts, even jupiter broadcasting also pointed it out that many users have moved away from Ubuntu, a lot of Tech Sites have also has been saying to use Mint>Ubuntu to not use Ubuntu at all

[leif81]

Fedora. Because it usually has the latest upstream development features first. IMO, it's the most exciting to watch.

[duby229]

Please include Gentoo, I'd like to see what real benefits you gain from compiling all of your own software.

I don't use it, but it's really the only distro that has a huge difference between the conventional ones.

I use gentoo everyday. Compiling everything doesnt make a huge difference in performance. It's pretty close. The truth is that most distro package maintainers (at least for the important packages) are reasonably competent and package fairly good binaries.

[LinuxGamer]

Fedora. Because it usually has the latest upstream development features first. IMO, it's the most exciting to watch.

This is what i was Thinking Fedora gets all the new Tech thing's like systemd wayland etc
Fedora 20 looks really good to

[phielix]

Do you realize LFS isn't even a distro, in the traditional meaning, do you? I mean, it's just docs, they don't even keep a central repo of sources (that's what keep it from being a source based distro). Manually built OS can't count as a distribution, because you are not actually *distributing*.
The idea of benchmarks is for users to get an idea of how it will run for them, and LFS is the furthest from something you can extrapolate, it's 100% custom.
Also, Debian unstable, maybe, but stable wouldn't help.

I'm aware of that and yes I meant debian stable! So lets set the two (debian stable as distro and LFS as non-distro) in brackets like this (debian and LFS) as recommended systems to be bench-marked just for reference! So everybody can see how a untouched system would perform in comparison to real distros!

That's the idea!

PS: LFS should be in a given preset of configuration that everyone can verify!

[mrugiero]

In the short term, yes. In the long term it may be too difficult for Lubuntu maintainers to keep both standalone X.org and LXDE on X.org running correctly.

Well, yes, when the toolkits drop X that might mean problems.

Fedora. Because it usually has the latest upstream development features first. IMO, it's the most exciting to watch.

Probably yes, but it seems by default they use debug build on bleeding edge packages, and this might lead to misleading results (everything would look slower than it would really be when it comes to end users).

[mrugiero]

I'm aware of that and yes I meant debian stable! So lets set the two (debian stable as distro and LFS as non-distro) in brackets like this (debian and LFS) as recommended systems to be bench-marked just for reference! So everybody can see how a untouched system would perform in comparison to real distros!

That's the idea!

PS: LFS should be in a given preset of configuration that everyone can verify!

So you meant as an extra benchmark? Thought you mean actually switching.
In that case, it might be used as a comparison between minimal system and distros.