First GNOME Census Results

supersloshy writes "The GNOME Census, a project to see who contributes to GNOME and how, has released its first set of results. The results group people by their reasons to contribute code, what they contributed code to, and what percentage of the total contributions they have. For example, 23.45% of code contributions were volunteer, 16.3% of code contributions came from Red Hat, 1% of contributions came from Canonical (which has caused a lot of controversy), and 0.24% came from Mozilla Corporation. The census results are also represented in diagrams (release activity, why contributions were made, and what was contributed to and by who). The report is also available here and is licensed under a Creative Commons Attribution Share-Alike license."

Comments

doesn't seem that scandalous

[Trepidity]

If the linked post is an accurate overview, at least, it looks like Red Hat is doing a lot more contributing to GNOME's core, while Canonical is doing a lot more building of apps, widgets, and other miscellaneous desktop stuff on top of GNOME. Both seem like reasonable things for an open-source company to contribute. Linux desktop environments need more hacking on the core, and need more interesting things built on top of that core too.

Re:doesn't seem that scandalous

[buchner.johannes]

Yeah? Link?

Maybe that coincided with the time when there were no free 3D drivers, and Red Hat forwarded the "Fuck You" from hardware vendors since they (a) hadn't the drivers developed yet and (b) had a principle problem with including proprietary drivers.

Then (in my opinion) Ubuntu et al not-so-strict distros included proprietory drivers, Linux became more present on the desktop, Hardware vendors noticed Linux. Open-source driver developers had more time and resources to continue and eventually brought forth free drivers.

What's your version?

Re:doesn't seem that scandalous

[crush]

Fedora and Red Hat provide Free Software in their repositories. It's trivial to install the non-Free drivers (and their associated hidden bugs) supplied by NVIDIA.

In addition to that Debian, Red Hat and Novell and Intel and other honest players have spent huge amounts of time coding up Free drivers with the Nouveau project (free NVIDIA drivers), Intel drivers, and ATI/AMD drivers

Sounds like the only one saying a big FUCK YOU is your self.

Re:doesn't seem that scandalous

[thetoadwarrior]

That's complete BS. Like the link pointed out, they may be trying to submit stuff and it's not accepted. But even if that isn't the case the one thing that Ubuntu does much much better than anyone else is provide a huge collection of useful easy to read documentation that can be applied to most any Linux distro.

They may not be doing much coding work but they are making people more aware of Linux and rather than fighting each other and making the community look like a bunch of dicks we should appreciate what they have done because quite frankly they didn't have to do anything and the competition is good. Fedora wasn't that great for working straight out of the box, from my experience a few years ago but Ubuntu was. That means users had a good option to try (better for everyone) and Redhat needed to improve which, I think they have.

If writing good code was all it took to be a popular OS then Linux should be a lot more popular than it is now. It's not and that's because it needs people doing other things some of which Canonical are doing.

Re:doesn't seem that scandalous

[Will.Woodhull]

Parent post, and the articles that are raising this "controversy", are comparing apples to oranges.

Red Hat and Canonical are both commercial entities. But Red Hat has been profitable for several years; Canonical has yet to generate a profit-- it is still in its start-up phase. One cannot expect a business that is still completely dependent on an angel's generosity for financing (thank you, Mark) to be as active in the community as a business that has a positive cash flow. Red Hat has the resources to pay some of its personnel to shepherd its developments through the upstream process. Canonical has chosen to put its limited resources to other tasks.

Also note that all the Ubuntu's distros are supported by Canonical. But Red Hat split off Fedora sometime before Ubuntu became a player. Red Hat now has no free-as-in-beer offerings that compare with the Ubuntu distros.

And in any case, Ubuntu's primary contribution is in meeting the needs of newcomers to Linux and FOSS, and it does that exceedingly well. Better than any other distro has done before. Proselytizing-- raising the public's awareness about Linux and FOSS-- is generally recognized as a valid mode of contributing to FOSS. One that is especially appropriate for individuals and businesses who are not in a position to contribute code. Ubuntu has done more in this area than Red Hat has ever done.

The spokesman from Red Hat who apparently set up this "controversy" should be ashamed of his words. A significant portion of Red Hat's new clients are from businesses whose managers tested the Linux waters with Ubuntu on their personal machines, and then went to Red Hat for its expertise in supporting enterprise systems. Ubuntu does not compete with Red Hat and is not riding Red Hat's coattails; Ubuntu's existence drives business to Red Hat.

Nota Bene: In 2008, Mark Shuttleworth guessed that Canonical might become "cash flow positive" in 3 to 5 years. The recession has probably pushed that forward somewhat.

Half the story

[ThoughtMonster]

The census is correct in implying that Canonical has not as many modules in upstream GNOME repositories, however that is only half the story. The census counts all commits since the beginning of the project, so Red Hat has a 6-year head start. Not to mention that Red Hat is a much bigger company than Canonical.

Canonical provides a lot of things of value to GNOME and the free software community in general. The (recently established) Canonical Design Team produces research on software usability, the value of which is not easily quantifiable. Many pieces of GNOME software live on Launchpad and are not strictly part of GNOME upstream (Simple Scan, for instance). This might change if (or when) these modules are accepted in GNOME proper.

To claim that Canonical is freeloading on other companies' contributions is a bit of myopic, in my opinion. How many upstream bug reports came from Ubuntu users?

Re:Half the story

[buchner.johannes]

To claim that Canonical is freeloading on other companies' contributions is a bit of myopic, in my opinion. How many upstream bug reports came from Ubuntu users?

Too many, we marked them as dup. But your point is invalid since Canonical != Ubuntu users and Canonical != Ubuntu maintainers. Latter are all in the volunteer camp. Red Hat users & maintainers are probably largely there too.

The way I see it Ubuntu is mainly a packager (distribution) and behaves like one. They mainly configure, build and distribute the existing software. Of course they provide patches for bugs they encounter, and they send it upstream to reduce their own work.

But Canonical doesn't have the means and will to truly commit developer resources to Linux (like Red Hat does). They want to achieve something with what is there*, and they are very good at communicating, community-building, reacting to users, connecting users and developers. That is Ubuntu's value.
Red Hat has some of this too, but for them it is business to engineer a Linux that works, because that is what they sell.

*Greg Kroah Hartman complained Ubuntu doesn't give patches upstream.

Re:Half the story

[ThoughtMonster]

... But your point is invalid since Canonical != Ubuntu users and Canonical != Ubuntu maintainers. Latter are all in the volunteer camp. ...

I disagree. In principle, you are correct, Canonical, as a company, has nothing to do with me, as a user, filing a bug report on some piece of software. However, how many of these bug reports would exist in the first place if not for Ubuntu, for which Canonical is largely (if not wholly) responsible? Something about eyeballs and shallow bugs.

For me, Canonical succeeded where most other companies did not, in marketing Linux and GNOME as user-friendly solutions, which in turn, I believe, will draw developers to produce more software for Linux.

Whilst this is, in part, due to the relative maturity of both products, for which Red Hat is largely responsible, I believe that GNOME benefits greatly from Canonical's approach towards user-friendliness as much as Canonical benefits from the infrastructure on which they base their products. Canonical has produced great software (like Upstart) which may not be obvious.

Re:Half the story

[houghi]

Greg Kroah Hartman complained Ubuntu doesn't give patches upstream.

http://video.google.com/videoplay?docid=3385088017824733336#
Keynote address by Greg K-H given during the inaugural Linux Plumbers Conference Sept 17, 2008 in Portland, OR.

Re:Half the story

[crush]

Many pieces of GNOME software live on Launchpad and are not strictly part of GNOME upstream (Simple Scan, for instance).

That's the problem: Canonical is not doing the hard work to get what little they do write upstream. Stuff that is not upstream is just balkanized, fractured, non-maintainable code. It doesn't provide any benefit to the rest of the GNU/Linux community, i.e. the people that write all the rest of the code and upstream it so that Canonical can exist in the first place. Usability research is useful, but when I click your link I see one study (on Empathy) and further clicking around on the Canonical Design team site reveals that, as so much of Canonical appears to be, it's all about marketing. Seriously: ONE study and then three guides devoted to "guidelines to support the brand documentation and help create consistent brand usage."?

Apples and Oranges

[WarJolt]

Redhat is publicly traded, has over 9 times as many employees as canonical and actually makes a lot more money then Canonical.
Quit picking on the small fry. Ubuntu contributes enough.

Re:Apples and Oranges

[JonJ]

The Ubuntu fanboys enjoy ripping on Red Hat for not contributing to the desktop, and for being a boring company focusing only on the serverside of things. Seems like that position is bullshit. People are also claiming that Red Hat doesn't care about the desktop, which this proves is also pure crap. For the people positioning Ubuntu as the desktop champion of GNU/Linux they're not contributing anywhere. Not the kernel, not GNOME, no where are they contributing a significant amount of patches. And all the apologists are busy trying to justify that the most downloaded distribution on distrowatch is not giving back anything significant. Even the fluendo guys are contributing more than Ubuntu.

Re:Apples and Oranges

[brejc8]

They've also made it easier to install proprietary drivers, which is always a mess in Fedora.

[citation required]
And not just a post by an Ubuntu user who heard it off a friend. I hear this every day and never met anyone who has supporting evidence. Along with "Fedora is just for servers", "Fedora uses bleeding edge so nothing works" and "there be dragons in them hills".

I install Ubuntu, Suse and Fedora on university machines on a daily basis. There is no massive discerning difference between these distributions that makes one much easier for 3d drivers than the other. All three have package repos for proprietary drivers and are as easy to set up.

Re:Apples and Oranges

[arose]

Is this over the lifetime of Gnome or that of Canonical. If it's the former, then your argument doesn't hold water. Yay for statistics without methodology!

Re:Apples and Oranges

[brejc8]

I install whichever OS they ask for. I also install Windows XP and getting the accelerated drivers for that is actually a pain. Why all three? Because there are Suse fans who feel most comfortable using Suse. Because there are multiprocessor simulators which are distributed only in .deb packages. Because, and I really must strongly emphasise this, there really isn't much of a difference between the distributions and there are no dragons!. They are all collections of the same software. If I solve a bug on one, it is the same solution on the others. This is why I feel strongly that distros should upstream their efforts. But perhaps most importantly, I do this because I want people to be comfortable using their computers. The reinstall cycle is about once every 2 years. I would appreciate you not insinuating people being mentally deficient on the ground that I put in more effort that is strictly necessary, after all the open source community is driven by people who put in more effort than the minimum necessary to get the job done, in order to make others' lives better

Re:Apples and Oranges

[Anonymous Coward]

Did you miss that it's a university? Logic need not apply, especially in turf wars where CS, Physics, and 3 different engineering departments all disagree on what OSes are acceptable, and IT has no authority to settle it because they're staff, not faculty. If you can get by on 2-3 Windows versions, Mac OS X, and 3 Linux distros, you're doing pretty well.

Languages

[maxwell demon]

One interesting observation about the contributions on language bindings: Obviously volunteers are mostly into scripting languages (Python, Perl), while each compiled language is dominated by a single company (C++ by Openismus, Java by Operation dynamics, and C# by Novell).

Freeloaders = good

[Alwin Henseler]

To claim that Canonical is freeloading on other companies' contributions is a bit of myopic, in my opinion.

'Freeloading' often has a negative meaning, but in open source land the opposite is true IMHO. Any additional user helps to improve the software just by using it:

• Increased user base means increased market share, bringing open source software closer to the point where companies take Linux support more serious for their products, governments may take a 2nd look at their open source use & support for open standards, websites are checked more often in alternative (read: non-IE) browsers, etc, etc.
• More users = more testers, more bug reports etc. This ultimately helps the software quality, if more bugs are found (& hopefully, fixed).
• More users = (over time) more experienced users, that can help newcomers get started.

So regardless of who deserves credits, that's many networks effects that benefit all users of such software, Gnome included. Freeriding on that is about as harmful as watching new years' fireworks without lighting any of your own - you still contribute to the party, just by being there. And in that sense, Canonical has done a lot to support Linux - by attracting & supporting many new users.

Re:Freeloaders = good

[noidentity]

This isn't even freeloading, because it doesn't put any load on a project to use a copy of its source code. This isn't something physical where it's limited.

Re:Freeloaders = good

[Group XVII]

Your argument made me feel nostalgic for the days when using Linux meant testing software and filing bug reports. Now I use Linux Mint (based on Ubuntu) and that world is forever lost to me. I am hooked on freeloading. Everything just works and I just let it.

Canonical's code contribution

[eddy_crim]

Canonical's code contribution is irrelevant. What open source has always needed is some polish and some marketing. Thats what canonical provide, they polished and marketed (to an extent) a decent distro. OSS has never been short of decent code and quality software engineering. Canonical are providing a great link in the value chain of linux and as long as the basic prinicipals are upheld im all for it!

Re:Canonical's code contribution

[Anonymous Coward]

take a look at various start pages of linux distributions. guess where you won't find the word "linux" even mentioned...

Re:Canonical's code contribution

[icebraining]

Debian GNU/Linux is a free distribution of the GNU/Linux operating system. It is maintained and updated through the work of many users who (...)

Re:Canonical's code contribution

[Anonymous Coward]

And this is exactly why it works. Do you see Google marketing Android as Linux based? Certainly not anywhere outside of it's dev community.

Linux, believe it or not, has a bit of a scary co-notation with the general public. Sure, it has come a long way from being called "for hackers only", but it's not there yet. Calling it Ubuntu and dropping the Linux reference is a GREAT idea. I would recommend it to anyone starting a new distro.

Cocoa-based-candy may sound great to you, but I'll just have my Chocolate thank you.

I call bullshit

[Anonymous Coward]

Except that as this report shows, Ubuntu has basically done jack shit to improve Linux usability.

This report doesn't show that. There are lots of code. Combining all that code to a package that is somewhat nice, well configured and works out of the box is damn important. Even more important is the massive amount of documentatiton that the Ubuntu community has created about pretty much everything. It would take an idiot not to recognize the value of those things. But let's take an example.

I have an extra computer in my house (but no extra monitors, etc.) and I decided to install it as a debian server. There was a slight complication, though: It is a bit noisy so I don't want to keep it in my bedroom and I also don't want to run cables all over the floors so I went to a shop to buy a wireless adapter. They are traditionally PITA on linux side so I went out to google for linux compatible devices first. I found an awesome list created by ubuntu community (didn't find anything comparable from anywhere else) and bought a device that worked out of the box on Ubuntu. I got home and tried to install it on Debian... Which I didn't succeed at. I found a guide, it had a number of broken driver links. Then I found more guides with more links. After an hour or so I was able to aquire the package... But I couldn't get past 'make'. Another hour trying to get past the problem for no avail. (for the record, I'm a sophomore year software engineering student going for bachelor's degree. And at some point I asked my roommate, sophomore going for CS degree in another university, for help). Then I gave up trying to do it that way, decided "Meh. I'll have to at least use stuff from Ubuntu repos anyway...", formatted debian, installed ubuntu and it all worked out of the box. As it always does with Ubuntu.

I haven't had much experience with red hat lately but the work that Canonical is doing is obviously valuable.

Essentially, Ubuntu is claiming that they've brought Linux to the desktop. Except they haven't. Red Hat has done more - 16 times as much, in fact - towards getting Linux on the desktop, but Canonical is taking all the credit for Red Hat's effort.

You may say that "As Red Hat has been around longer and contributed constantly the whole time and done a lot of marketing, etc... Their overall contribution to Linux exceeds Canonical's by a wide margin" and I would completely agree with that. But if you say "Red hat has contributed 16 times as much code == red had has contributes 16 times as much to bringing linux to desktop", you are very, very wrong.

Re:I call bullshit

[icebraining]

My guess: you used stable Debian.

Drivers usually come with the kernel, and Debian trades new drivers and other features for a stable environment. Ubuntu simply ships a more recent kernel, at the expense of less testing.

Another option is simply using Debian Stable with a backported kernel. It's as easy to install as normal Debian, but comes with a more recent & less tested kernel.

Re:I call bullshit

[crush]

1. You're conflating Ubuntu and Canonical.

2. Canonical is a large, private company which has been around since 2004. If we compare the contributions only since 2004 then Red Hat has still contributed more code than Canonical: to EVERY part of the Linux stack. More egregiously if we compare the large, well-funded Canonical to small start ups like Litl, Collabora and Fluendo even then Canonical fails to contribute as much.

We've come a long way since our launch in 2004. We now have over 350 staff in more than 30 countries, and offices in London, Boston, Taipei, Montreal and the Isle of Man.

Everyone puts these Canonical freeloaders to shame.

You would indeed be wrong if you merely said "Red Hat contributes 16 times as much code". That's ONLY what they contribute to GNOME specifically. They develop the kernel, most of the toolchain for compilation, vast parts of the network stack, fonts, ... basically bloody everything AND they do that by adhering to Free Software and SHARING EVERYTHING UPSTREAM where it's easy for any distro to benefit from their work.

Re:I call bullshit

[at_slashdot]

Freeloaders? I think you don't know what free software is all about...

Re:I call bullshit

[Pollardito]

he should clarify whether he means freeasinbeerloaders or freeasinfreedomloaders

Re:I call bullshit

[Urza9814]

I found an awesome list created by ubuntu community (didn't find anything comparable from anywhere else)

Not sure why you couldn't find it, but Mandriva has had a database of supported hardware since before I started using Linux (which was Mandrake 9.2, in 2003). Having a list of supported hardware certainly isn't a new idea.
http://www.mandriva.com/hardware/

installed ubuntu and it all worked out of the box. As it always does with Ubuntu.

Glad you've had good luck with it. Last couple times I've tried I couldn't even get the installer to boot. And when my brother tried it took 4 days to get his wifi card to work (a card which works out of the box with Mandriva.) Stuck with Mandriva for many years because of that, though I've recently switched to Arch. And while it takes a couple hours to get the system setup initially on Arch, I couldn't be happier. Haven't had a single problem since installing it.

Re:I call bullshit

[siride]

Kernel modules don't matter. They are just a way of breaking up the loading process. Once a module is loaded, the code it contains is practically indistinguishable from compiled in code. That's why modules can be built-in instead of compiled as modules.

Lines of code isn't the only thing that counts

[mattbee]

When it comes to bugs and usability problems, Ubuntu run a much sharper bug tracker - it usually has coverage of almost any minor GNOME issue. Between Canonical and their users, It might have taken many man-hours to track down, discuss and identify a small usability bug, which might only result in a fix of a few lines of code. It's not about turning the screw, it's knowing which screw to turn. So counting lines of code as the only contribution is completely unfair to Canonical.

This doesn't just go for GNOME; the best discussion of kernel and firefox bugs usually ends up being hosted on Ubuntu, just because they have fostered the largest community of enthusiastic Linux desktop users.

Re:Lines of code isn't the only thing that counts

[Spewns]

None of them would ever consider investigating bugs, talking to people upstream, downloading the code, submitting patches.

They're users. Most simply don't have the know-how to do these things, and it's unreasonable to expect them to, especially in the "downloading code and submitting patches" department. You make it sound like a trivial thing for even hobbyist programmers to do, especially with the bloated, ad hoc codebases they'd probably be dealing with. And if you're hearing from them, they are talking to upstream. Upstream to Ubuntu users is the Ubuntu forums or Canonical.

Re:Lines of code isn't the only thing that counts

[TheLink]

> This is not contributing back to the community.

I think a lot are pointing out which dishes suck. They may not be able to tell you exactly why and how, but if your target is the general public, that's useful info if you know how to sort it out properly.

> All this noise distracts from the real contributors who actually do the work, quietly, productively and without much of a fanfare.

If the bug reports are distracting the workers then it's the fault of the organization.

The bug reports do not have to go straight to the developers. They can go to someone else first whose job is to figure out which are the top problems to be fixed - there are always bugs so you have to prioritize. Maybe someone could also figure out whether the problem is a bug that's best fixed in a module or one that's best fixed by changing the architecture in the future - too often if people are too busy fixing stuff at the tree level, they don't fix stuff at the forest level.

If you want to get paid..

[onion2k]

If you want to get paid for what you do then charge for it. I don't mean money necessarily. There are lots of ways of getting paid. But charge something.

In this case the reciprocal amount of work people are demanding from Canonical is a form of payment. If you want to claim it's not "fair" that one company is doing more for a project than another you've got to set up the system to stop them, otherwise you have no grounds for your complaint. You can't set up a stall with a big sign saying "Free, please take what you want, no need to give anything back in return" and then moan when someone takes you up on your offer.

Re:If you want to get paid..

[blackest_k]

I wasn't going to post but maybe this is the reason for the tribal comments.
There is no value in measuring contributions by various sections of the linux community commercial or others.

Red hat makes valuable contributions so does Cannonical, so do many other companies other than SCO i think we can agree on that. I have a bias towards ubuntu it works on my systems and i am familiar with it. However i don't think the sun shines out of cannonicals arse , that now belongs to oracle.

but seriously measuring contributions made by other people is divisive and unnecessary. I've contributed in a few area's and just helping newbies is a contribution that most can make. The size of the contribution doesn't matter. Redhat is commercially successful and turns a profit mark shuttleworth pumps money into Linux via ubuntu and the parent seems to say that isn't good enough do more.

really size of contribution is up to the contributer and assigning value to each contribution is just divisive
linux is being used more by more people and there are more people contributing thats all we need to know.

getting partisan and making digs only makes for trouble.

all contributions have some value

Check which modules get rejected

[Anonymous Coward]

The reason why RedHat's piece of GNOME commits is so big is because they have been rejecting modules developed by competing companies. Novell made a push to get their start menu included in GNOME, it was rejected by the RedHat majority. Same thing with Compiz, a compositing window manager developed by David Reeveman of Novell, also rejected despite it being an almost complete drop in replacement for Metacity which is ancient RedHat technology. He also worked on bringing OpenGL into xorg and had a working prototype for how to do it. Also rejected because RedHat favored a different approach by writing AIGLX. The reason why Novell doesn't have a large stake in GNOME's codebase is certainly not for a lack of trying. There are dozen more modules that have been rejected over the years. What they all have in common is that RedHat employers aren't working on them.

Then check what modules have had no problem getting included: PulseAudio, Clutter, DeviceKit, Cheese, gnome-user-share... All created by RedHat employers. Basically, when it comes to the core of GNOME's infrastructure, RedHat has been very effective in keeping outsiders out.

Re:Check which modules get rejected

[Bazer]

Let's recap what both companies contributed:

• Novell: Compiz, XGL (unmaintaned X server with OpenGL)
• Red Hat: PulseAudio, Clutter, DeviceKit, Cheese, gnome-user-share

Notice something about the scopes of each of those projects?

Same thing with Compiz, a compositing window manager developed by David Reeveman of Novell, also rejected despite it being an almost complete drop in replacement for Metacity which is ancient RedHat technology.

Metacity ancient? What do you make of the whole X server then? Should we replace it too? Don't get me wrong. I don't dismiss Compiz as eye-candy because it's far more than that. It came way to early. It was unusable without proprietary drivers and unstable with. To this day Compiz has problems with stability on anything but maybe Intel boards. The necessary groundwork just isn't there yet.

Re:Check which modules get rejected

[Anonymous Coward]

Novells menu for GNOME is crap! I just used it again on my mothers Netbook and you actually can not use it at all. When ever you want to launch a not-so-often-used application program. It is faster to launch gnome-terminal and start it from there than go "all programs" -list.
The "favorites" is such that you have difficulties to get it ordered as you want. Or to get all wanted favorites there but still keep the most used list big enough, even with 1920x1200 resolution. KickOff is much better what Novell did but they could not manage get it done for GNOME.

And what comes to Compiz-Fusion, it is crap! It is already dying (what some contributors are saying) and it is too complex code to maintain. And I think you did not know that Compiz as wrote behind the doors so no one could contribute and make it better. No, they wanted to have in done closed way.

PulseAudio has got just too bad reputition, mainly because Ubuntu users because Canonical did not configure it propely. I have used PulseAudio almost since it was published in main distributions and it has worker perfectly. Only thing what I can whine is that it has only GTK+ configuration frontends and not Qt for KDE SC.

Clutter is owned by Intel, not by RedHat. And it was developed by other company before Intel bought it. DeviceKit is neither RedHats work, it was just taken in use first by Fedora, then some other distros and after that the Ubuntu. So do not whine about it either.
Cheese is done by one guy (Daniel G.Siegel) and purely for GNOME upstream. So you miss again. Seems you just want to blame RedHat for that Canonical just has not wanted to contribute to the upstream.

And if you have not understanded, RedHat does not control GNOME. GNOME is part of GNU. Unless you want to say that GNU (and so on GNOME) does not control itself, but is controlled by RedHat, then you would be correct.

But please, just stop building wild theories how RedHat is the evil, just like Canonical is calling RedHat as propietary company while it itself is such!

Re:Check which modules get rejected

[drinkypoo]

Notice something about the scopes of each of those projects?

I notice that Clutter is an intel project. I notice that Cheese is crashy and slow. I notice that Xgl died in favor of a solution which still isn't here, and that more of the system was composited when we had it. In fact, I notice that Compiz window effects (since you bring it up) worked probably three times faster/smoother under Xgl than they do under the modern AIGLX desktop.

Metacity ancient? What do you make of the whole X server then? Should we replace it too?

Large parts of it have been replaced already. Xgl provided DRAMATIC performance improvements, I know, because I've actually run it. I was pretty upset when they killed it since I STILL can't get performance that good.

To this day Compiz has problems with stability on anything but maybe Intel boards.

Uh, what? I've recently (say, in the last year) had metacity die just about as much as emerald, which is less stable than the default gtk-window-decorator that looks just like metacity.

I think you're a troll utilizing selective memory to try to support a point.

16 times? Strange metric...

[mangu]

Red Hat has done more - 16 times as much, in fact - towards getting Linux on the desktop

That's assuming each line of code has the same value.

For me the important point is with which system can I get a computer working quicker and with less effort for installation and maintenance. Ubuntu wins.

OK, you may say that this only reflects the superiority of APT over RPM. Comparing Ubuntu with Debian, Ubuntu wins again.
 

Re:16 times? Strange metric...

[MSG]

For me the important point is with which system can I get a computer working quicker and with less effort for installation and maintenance. Ubuntu wins.

Ubuntu wins for you. I suspect this is primarily related to your greater experience with Ubuntu. I, on the other hand, can set up a RHEL system much faster and with less effort for installation and maintenance than I can with Ubuntu. RHEL wins for me. Having done both, I can state objectively that there are fewer steps to get a working RHEL web server (for instance) than there are for Ubuntu.

you may say that this only reflects the superiority of APT over RPM

Well, compare apt to yum since those are more similar tools. Apt is faster, I'll grant you. Yum, on the other hand can install a local package and resolve dependencies from the repositories and it can install a package given the path of a file it provides. Apt cannot do those things, so I believe yum to be the superior tool.

Re:16 times? Strange metric...

[hdparm]

Oh, FFS.

Superiority of APT over RPM? Get a clue. You can compare APT and YUM and how well they manage whatever packages your distro of choice have.

Fedora 13 installs everything I need for the laptop out of the box - wireless driver, mobile modem driver, even bloody compiz works on ATI mobility card without any additional requirements. YUM is rock solid for ages now. The only extra thing needed is rpmfusion repos to get proprietary codecs going.

Open source

[nxsty]

Greg DeKoenigsberg, an ex-Red Hat employee wrote a blog post slamming Canonical for the "absolutely egregious" statistic and suggesting that Canonical has been "riding on Red Hat's coattails for years." Tough shit. This is open source, if you don't like others using your work you should develop proprietary software instead.

commits aren't sufficiently granular for comparing

[Artifex]

You can commit on simple edits (like a ui typo change), or on adding a whole new chunk of code.
And, if you're busy, you might make both an edit and the addition in the same commit.

Is someone who makes five typo change commits doing five times the work of someone adding one with a new function?
I seriously doubt it.

What matters is end user convenience

[shutdown -p now]

As a user, I don't care in the slightest who committed more patches, or lines of code.

What I do care about is how easy and convenient it is to use a particular distro. And there Ubuntu offers a lot. Try to play an MP3 file? Fail on Fedora out of the box; with Ubuntu, you get a dialog asking you if it's okay to download the codec - a single click, a brief wait, and it Just Works.

Or take drivers. As soon as it boots, Ubuntu prompts me to let it install proprietary NVidia drivers. A single click, and I have a 3D enabled system which actually works and has performance decent enough for gaming. Fedora? Either join the bug hunt with noveau, or search for a 3rd-party repository providing what you want.

Yeah, yeah, I know, Free Software is supposed to good for your karma, and friends don't let friends use proprietary crap. And Red Hat are your friends, right?..

... does anyone actually care?

Well, I guess some people do, and those people stick to Fedora. Judging by the amount of users it has compared to Ubuntu (and other distros who don't shove "FOSS only" into their users' throats), it's not as popular as some people would like it to be.

The linked blog post by an ex-RedHatter is dripping with venom over how Ubuntu "beats everyone at marketing", but totally misses the point. Ubuntu beats everyone at convenience and "just working" first and foremost; marketing is just icing on that cake. You want to make a principled stand over FOSS? Fine, but then don't complain when users flock elsewhere!

More numbers

[OneAhead]

Here's the original presentation of the gnome census. Most of the numbers discussed here come from slide 16 and 18.
http://www.slideshare.net/nearyd/gnome-census

I also don't agree with the claim that only 23.45% of contributions come from volunteers. There is also the 16.94% "unknown". Now, if you're working on Gnome for a company, you usually would want to list your affiliation. If you don't, maybe you're contributing to Gnome on company time without your bosses knowing, but such a situation should (arguably) be counted as "volunteer" work. But I speculate most of these unknowns are simply actual volunteers, who just skipped the question "which company do you work for", and didn't notice there was an option "none" (people usually are in a hurry when they fill in surveys). That would set an upper limit of 23.45%+16.94%=40.39% on the fraction of volunteers, and I feel the actual number is likely to be closer to this upper limit than to the lower limit (23.45%).

Ironic ...

[kbahey]

I find it ironic that Redhat are the ones complaining about Ubuntu, while it was Redhat who exited the desktop market years ago, focusing on the server side of things. This void that was created was filled by Ubuntu, and it has become successful. Fedora is not quite the same, since it is bleeding edge, with not stable releases.

Ubuntu's success is well deserved. They fill a much needed part in the Linux arena.

Counting patches from before Canonical existed is inaccurate and biased. And patches are not the only measure. There is packages, polish, community building and marketing.

Re:Ironic ...

[Arimus]

WTF?

RHEL has a workstation and server variant amongst the mirad of options, Redhat still provide alot of support to Fedora's efforts.

An integrated distribution is a contribution

[FoolishOwl]

There are almost thirty thousand packages available from Canonical's repositories. Assembling a coherent, working Linux distribution from a selection of available packages is, in itself, a massive work of engineering. Given that Ubuntu is the most popular Linux distribution for desktops, there's a strong case that they've designed the best available distribution for that niche.

If Canonical had contributed no software at all to GNOME, they would still be making a significant contribution to the free software ecosystem.

Re:What matters is end user convenience (offtopic)

[jyx]

As a user, I don't care in the slightest who committed more patches, or lines of code.

... does anyone actually care?

Actually, you should care. Just as you should care where and how your cheap shoes were made and what ingredients went into that chicken nugget you ate.

Your only power as a consumer is that of choice, and by being uninformed you cant make a more meaningful choice than 'this one looks pretty and is the cheapest'.

By being even slightly informed consumers *might* stop buying drm enabled music, they *might* by more ecologically sustainable products and *might* start to realise that the reason why they cant get component video from their HDMI sources is purely by the whim of the MPAA and start looking for alternatives.

Caring doesn't mean entirely hanging your way of life to be the perfect zero carbon imprint, 100% sustainable , super politicly correct guiding social light mega person. Caring just means that you having an interest *why* things are instead of blindly taking the path of least resistance.

Slightly on topic, you are correct about Ubuntu's user friendliness, and a lot of software developers (closed AND free) should try and make the end users experience as painless as possible. But remember, it could be argued that Ubuntu could only do this because of all the amazing work that went into Gnome in the first place.