User: Password:
Subscribe / Log in / New account

Notes from the Ubuntu Developer Summit (The H)

The H rounds up reports from the Ubuntu Developer Summit that is currently being held in Oakland, California. Chris Kenyon, Canonical's Vice President of OEM Services, reports that Ubuntu shipped pre-installed on 8-10 million computers last year and predicted that it would ship on 18 million next year (which would be 5% of the market, he said). Also: "Ubuntu developers are planning to fork the GNOME Control Center to create their own Ubuntu Control Center package. Other than GNOME Shell, it is planned that the installation CD for Ubuntu 12.10 "Quantal Quetzal" will include almost all core components of GNOME 3.6, including Clutter. Up to now, Clutter has been missing from the default install which had forced the Ubuntu developers to include Totem 3.0 instead of 3.4 because the newer version depends on Clutter."
(Log in to post comments)

Notes from the Ubuntu Developer Summit (The H)

Posted May 10, 2012 17:16 UTC (Thu) by littlesandra88 (guest, #64017) [Link]

I am not quite sure what Clutter is. Is it a window manager?

Notes from the Ubuntu Developer Summit (The H)

Posted May 10, 2012 17:25 UTC (Thu) by BlueLightning (subscriber, #38978) [Link]

Notes from the Ubuntu Developer Summit (The H)

Posted May 10, 2012 17:29 UTC (Thu) by littlesandra88 (guest, #64017) [Link]

Yes, it says Clutter is a library, which completely messed me up, and what trickered me to ask.

Notes from the Ubuntu Developer Summit (The H)

Posted May 10, 2012 19:09 UTC (Thu) by rillian (subscriber, #11344) [Link]

Clutter is more like an animation framework on top of OpenGL. It makes it easy to write slide-y, fade-y visualizations or user interface components. Might be a more helpful link.

Notes from the Ubuntu Developer Summit (The H)

Posted May 11, 2012 10:36 UTC (Fri) by dgm (subscriber, #49227) [Link]

Maybe Mutter is confusing you. Mutter is the Gnome 3 window manager, and it's based on Clutter.

Trustworthy sources...

Posted May 10, 2012 20:25 UTC (Thu) by sebas (subscriber, #51660) [Link]

Even from my remote arm chair, I can tell that the article is in large parts inaccurate. Probably not such as good idea to use Phoronix as a trustworthy source, let alone their forums. The Phoronix-forum poster seems to be large uninformed about unity-2d contributions:

As to forking GNOME control center, it seems working with upstream is largely out of fashion, there's not a single sign that Canonical even tried to collaborate with GNOME. I've come to expect very little of that peculiar relationship, but surely people can do better than *that*?

*Creeps back into armchair*

Trustworthy sources...

Posted May 10, 2012 20:46 UTC (Thu) by littlesandra88 (guest, #64017) [Link]

Yes, I have stopped reading anything that comes from that sites a long time ago.

Michael will write anything to get clicks and visits. It is so sad =(

Just take his (in)-famous kernel bug article =)

Trustworthy sources...

Posted May 10, 2012 21:17 UTC (Thu) by ovitters (subscriber, #27950) [Link]

Canonical likes to do their own thing. They do try to upstream their designs every so often, but their focus is Ubuntu. So the timing is not always right (e.g. code drop while GNOME is in various freezes), or no time to incorporate feedback from upstream.

I don't really see a problem with a temporary fork, but IMO that should be temporary. Also, things should be done carefully. E.g. gnome-control-center in Ubuntu 12.04 is already not doing the right thing for gnome-shell in 12.04 (they changed it to use gconf again, while gnome-shell looks in gsettings/dconf; see

Differences like that is painful for upstream. Might result in upstream developers totally ignoring the bugreports from that particular downstream (just not worth the effort to figure out if it really is upstream, or another downstream change). Though as said, there can be value in temporarily trying something out (forking).

Trustworthy sources...

Posted May 11, 2012 3:58 UTC (Fri) by michaeljt (subscriber, #39183) [Link]

>E.g. gnome-control-center in Ubuntu 12.04 is already not doing the right thing for gnome-shell in 12.04 (they changed it to use gconf again, while gnome-shell looks in gsettings/dconf; see

I suppose my immediate thought when looking at the bug report is that GNOME is not going out of their way to make things easy for out-of-tree users either (Linux kernel style?). However I don't really know much about this and may be completely wrong.

Trustworthy sources...

Posted May 11, 2012 7:23 UTC (Fri) by ovitters (subscriber, #27950) [Link]

You are completely wrong.

I don't even know what you're after. Already stated that nothing is wrong with forking, but, if you fork, it makes things difficult.

Expecting upstream to take responsibility for the pain of forking: I totally disagree. Nothing against forking, but a certain pain is involved, and saying the pain should be relieved by upstream: terribly weird statement. Comes across as a "doctor it hurts when I do this; why didn't you do something to avoid it?".

But this case is not about forking at all. The issue is that apparently Compiz looks at the same gconf keys as metacity and gnome-shell used to.

When Metacity and GNOME shell were ported to gsettings, an announcement was made about this. I think it was seen (as breakage was known), but it doesn't really matter. If you change things, don't rely on upstream to avoid any negative impact of that change.

Note that I'm against purposefully making forking difficult.

Trustworthy sources...

Posted May 11, 2012 7:34 UTC (Fri) by michaeljt (subscriber, #39183) [Link]

Perhaps we are misunderstanding each other. From the bug report you linked to, I gathered that the reason for the change is because Unity is still using gconf, which GNOME no longer wish to support. This lack of support is what I meant when I talked about GNOME "not going out of their way to make things easy for out-of-tree users". And time for a statement of my own. While I definitely prefer projects to provide this sort of support I am not criticising GNOME either, as I can't see anything which gives me a right to tell them how to run their own project. Though I very much wish there was some fair-to-the-developers way of making FLOSS projects more accountable to their users (I know there are a few things, but as yet there is no general recipe).

Trustworthy sources...

Posted May 11, 2012 9:16 UTC (Fri) by paulj (subscriber, #341) [Link]

Though I very much wish there was some fair-to-the-developers way of making FLOSS projects more accountable to their users (I know there are a few things, but as yet there is no general recipe).

There is a very well established way, across many spheres of human activity, to make producers accountable to consumers: by having some equal exchange of consideration and responsibilities between the 2 parties. E.g., the consumer pays money to the producer in return for the producer taking some responsibility. For software, you could call it a support contract. ;)

The problems in the desktop Linux world here are many-fold:

  • Few desktop users have an expectation that they should pay for their Free Software. Some use Free Software perhaps primarily because of its 0-cost. Some users may even be happily paying for OS updates on other platforms (e.g. OS-X), funding the competition, or applications (markets). So the Linux desktop vendors are starved for resources, even while quite a few of their customers are happily funding Linux competitors.
  • For those desktop users who would like to pay, there is terrible fragmentation amongst the producers. Worse, the largest producer - employer of a great many Linux desktop developers - does not provide any way at all to pay it for support on a desktop that isn't hideously out-of-date for most parts of each decade. Its desktop distribution its developers work on is bleeding edge R&D, and its goal is to advance the state of the art - not necessarily to keep its existing userbase happy. The determined desktop user could perhaps take out a support contract with a number of different Linux distribution vendors, to cover all the bases, but that could still lead to problems and/or a lot of extra work if ever they needed to actually avail of that support, and would be useless for packages that are in the R&D distribution (e.g. there's no RHEL desktop with gnome-shell).
  • The packaging and ABI fragmentation across all the distributions, that make it hideous for 3rd parties to provide software for Linux. Other threads here recently have been discussing this. This surely depresses the software ecosystem around Linux, which depresses the size of the user-base desktop Linux can have, which depresses the level of support the producers can offer, and around in a loop. Generally, it means there's less resources available to develop the Linux desktop.
  • The "Well, if you don't like it, go fix it" attitude. Many users are incapable of fixing their problems. Further, while some users may have the technical capability, they may not have the resources (e.g. time) to spend on anything but the odd bug, very irregularly. It doesn't scale to have every end-user have to become an expert in every piece of software they might have a problem in. The attitude should be "If you don't like, pay someone to fix it" (which could be yourself). The "Have you filed a bug?" response to many complaints is also unproductive. As if the problem with Linux is that all the bugzillas are empty, and the armies of engineers are sitting idle, tapping their fingers, for want of a user to open a bug.
The problem with desktop Linux is that there aren't enough engineering resources to provide a polished, well-supported (today), end-user focused desktop experience. It's very very close, and it's amazing what's being accomplished by the developers and engineers there are today! However, it needs a bit more.

That there aren't enough resources for that is because there is massive FAIL in the Linux desktop world at extracting those resources from end-users and channelling them into the development, marketing & support of such a desktop. The reason for this is partly due to flawed expectations and attitudes in end-users and developers, it's also due to the balkanisation in Linux and lack of anyone able to do something to solve this; as well as a terrible lack of business fore-sight.

The lack of business foresight by one of the largest producers is particularly grievous. It seems to have been pulled in by the gravity of Suns' old customer-base and seems to want to follow its path - withdrawing from the end-user desktop to suckle on the teat of the enterprise server cash-cow. That strategy hasn't worked well in the long-term for any one else who tried it. Worse though, it means there's a goood number of their desktop Linux developers who are even further removed, resource-dependency-wise, from their users. No doubt, some number of them are not entirely happy about that situation either.

Trustworthy sources...

Posted May 17, 2012 9:29 UTC (Thu) by misc (subscriber, #73730) [Link]

Most distributions backed by a company do let you pay for having a professionnal version, around 50 to 100 euros per year, for a workstation. That's not something that is marketed that much, because almost no one target users directly, expect hardware makers. Microsoft sells windows to OEM, Apple is selling hardware, etc.

The cost of marketing to users is the same than with a company but the deal are smaller, so that's not as interesting to do.

Now, I think there is a huge misunderstanding on why there is free software. This is in no way to serve the wish of a consumer market, this is because science is faster when we collaborate.
And so the issue is that people mostly all use the R&D version, ie distribution lke Fedora, Ubuntu, Opensuse, instead of supported and stable release like RHEL or SLES. People do this because they prefer to have real fresh software, because this is easier to manage, because this is cheaper.

So if people prefer cheaper ( ie, 0 cost ) to paying, the only way they pay others OS is because they usually have no choice. And fixing this would be against free software spirit, IMHO.

Having more software could be done, but there is always more unsupported options that supported one, cause this is easier to not support. Again, that would be difficult to fix without a tight gripe like Apple has on Iphone.

The same goes for having a ABI, this requires someone to step and say "this should be done like this". Posix tried that, and it was not really great. FHS/LSB tried that, and this was not that great. Debian common core tried that, and it didn't worked. Heck, even the BSD achieve that, and people do not use them. So maybe we should just face that it move too fast for being standarized, unless there is 1 single person in charge of that. And that whatever efforts would be done, it wil be trumped by the fast and continuous innovation that is spawned by free software.

And asking to fill bug tracker is the same than when you go at your phone provider shop, asking for some changes on your phone, and being answered "please call this support number, I cannot do anything" ( happened to me to unlock my phone ). No one would say this is detrimental to cell phone market, and yet it appear.

It is the same when you go speak with some microsoft developpers, they will likely tell "please open a support case". So why do people accept this when they pay, and not when they do not pay ?

Trustworthy sources...

Posted May 18, 2012 16:42 UTC (Fri) by mpr22 (subscriber, #60784) [Link]

I thought getting a mobile phone unlocked mostly involved walking into a slightly dubious-looking storefront, handing the phone and a portrait of Her Majesty to the slightly dubious-looking gentleman behind the counter, and waiting a couple of minutes while he performs some arcana over the former and puts the latter in his pocket.

Trustworthy sources...

Posted May 11, 2012 10:21 UTC (Fri) by ovitters (subscriber, #27950) [Link]

Maintenance of gconf is somewhat of a seperate thing.

GConf is part of the GNOME 2.x platform, not the 3.x one. Despite that, it is still available + works (builds, etc) + used in 3.x. We already started porting from gconf to gsettings during GNOME 2.x (status at Various bits have been migrated before 3.0 and various after 3.0. The intention was that GNOME 3.0 would only use gconf. Unfortunately that wasn't feasable.

It is not possible is to have one component look at gconf and gsettings/dconf at the same time (for the same setting). That's also really weird to aim for IMO. Whenever possible, settings are migrated from gconf to gsettings/dconf.

GConf even during 2.x hasn't been actively maintained for many years. Even at the time that a lot of programs used it.

For every major release we made promises about what would be supported (API stability for gconf during 2.x). GConf deprecation happened various *years* ago. The result (things being ported) has been pretty gradual.

Trustworthy sources...

Posted May 11, 2012 17:42 UTC (Fri) by sciurus (subscriber, #58832) [Link]

I think you meant to say "The intention was that GNOME 3.0 would only use gsettings".

Copyright © 2012, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds