Shuttleworth: "Apple is driving the innovation"

    3 Postings

    Sees possibility for a Qt-based GNOME - An interview about the future of the Linux desktop and problems with "Hardy Heron"

    Last April, Ubuntu 8.04, Codename "Hardy Heron"was released, promising Long Term support - three years on the desktop, five years on the server. At the recent GNOME Users and Developers Conference (GUADEC) in Istanbul Andreas Proschofsky had the chance to sit down with Ubuntu-founder Mark Shuttleworth to talk about the new release, but also about the Linux desktop as a whole, the strengths of Apple and possible major changes to the GNOME platform.


    The following interview is also available in a german translation. A few months ago you released a new Long Term Support Release, how are going to maintain that further?

    Mark Shuttleworth: We just put out the first point release. One of the changes that we made with this LTS is that we decided we would put out a point release every six months, interleaved with our normal releases. So this gives us the opportunity to add new hardware support, to fix issues that are causing users distress. Speaking about hardware support, does that mean that you are also going to do kernel upgrades in between?

    Shuttleworth: On the kernel side - at least for the next two years - we are going to improve hardware support by adding to the current LTS-kernel. After that we might make it possible to switch to a newer kernel, I guess it depends on the extent to which it remains possible to backport drivers and so on. Ubuntu 8.04 had to take a fair amount of criticism cause of problems with Audio and other bugs, in hindsight: Should you have taken some extra time to fix those bugs, like you did with "Dapper Drake", the first LTS?

    Shuttleworth: Actually we based that decision to go with the six month schedule on what we learned from Dapper Drake. With Dapper Drake we thought that the extra two months hadn't added a tremendous amount of extra stability.

    The main criticism that I've seen for Hardy were first that we shipped Firefox 3 which was a beta. That was a very conscious decision taken in partnership with Mozilla and we were very confident that Mozilla in fact would release Firefox 3 in a reasonable amount of time. And if now - after the release of Firefox 3 - we would only have Firefox 2 on the desktop for three years, people would be equally upset. So I think it was the right decision.

    Another area were we got a lot of criticism was that we made a LTS release based on a GNOME release which had a very substantial change in the virtual filesystem layer. Most of this problems are fixed with the point release though, at least all the major issues have been resolved now, we've been working very closely with GNOME to achieve that. But that was not an ideal situation.

    The third you mentioned was audio and that is more challenging cause we see there is a need to get clarity on the audio stack in the Linux space. There are a lot of different permutations and combinations and we figured Hardy could go a long way to drive towards a consensus platform. So we met with other distributions and they seemed to want to make the same choices so we thought: Let's do that and send a very strong signal, that this is the audio platform that we think distros will embrace. Still it looked like you didn't pick a very good point of time for a new LTS: GNOME did big changes, Firefox was not ready at that point, you integrated PulseAudio, so wouldn't it have been better to use the next release, six months later, as LTS?

    Shuttleworth: I take that criticism. Though I think that all of that decision aligned us more closely with where upstream wants to be. So my sense is, we really shipped an LTS-release which is maintainable for three years. We knew there were some regressions but we felt we could address them during the maintenance cycle. So should people wanting a really stable Ubuntu wait for the first point release, like lots of Windows users are waiting for the first Service pack of a new release?

    Shuttleworth: It really shouldn't be like that, a release should be deployable. And we've no questions about whether or not that release is deployable, there are some questions whether or not very large scale infrastructures should immediately jump on the new release, which is a slight different question.

    So we didn't see a sufficient level of beta-testing during the test-period and many bugs are only filed when the release has already been made. So one option that we considered was: "Let's not call 8.04 the LTS, let's call 8.04.1 the LTS", so many people would upgrade who wouldn't use a beta and you get better feedback. So that's something we might do differently with the next LTS. So what about pushing harder to align the big distributions and get upstream projects to follow that cycle?

    Shuttleworth: Well you know that's a big push for me, I very much want to achieve that. In fact we said that we'll do an LTS in two years time unless we get an agreement with one of the other major distributions to coordinate. It's been very interesting discussions since I published this idea of having a meta release-cycle. Actually it's more about freezing at the same time than releasing at the same time, we really want to collaborate on a basic set of components. We already do a lot of work with Debian on that and we do well with some upstream projects, if we can achieve it across multiple distros that would be very significant. Do you see a real chance for that happening?

    Shuttleworth: Yes. I think it's quite feasible that one of the other major distros is making a limited commitment to that. There are some talks on this.

    There's a lot of precedence in economics that some level of coordination helps draw more customers, more users to everybody. There is a fear I think amongst the distributions that if we release at the same time as them users will be able to choose Apples to Apples. But the reality is, each of the distributions has a different set of values and that has nothing to do with which version of the are using. Still both Red Hat and Novell are quite bigger companies than you, having more support staff, more developers, so from their point of view, they might have something to loose, cause they are then providing more stability for you.

    Shuttleworth: Well we have a better security track record than Red Hat, we do that by focusing very hard on security, making sure the updates are available as fast as possible on Ubuntu, independent studies have generally ranked Ubuntu number one. We had a horrible security issue recently [the severe OpenSSL-bug recently discovered in Debian-based distributions, apo] but the response to that was exceptional.

    So what I'm trying to say here, that the notion that Canonical wouldn't contribute anything in such a situation and it would be a one way flow is something I disagree with. Look for example at the fact that Ubuntu has usually better hardware support, if we all were on the same kernel the others could take the drivers we put in there and have hardware support that is just as good as Ubuntu.


  Speaking about security, are you going to stay with AppArmor now that Novell has dropped the ball on it or are you going to change over to SELinux?

    Shuttleworth: That's a very interesting question. There's certainly been a lot of change there. But that's one of the tricky things when managing the evolution of a platform, you are taking a leap of faith on certain communities. Especially with communities which are largely driven by other companies there's a real risk with that, cause they might just send their strategy. It looks like everyone has chosen GNOME as their default desktop, do you think that observation is true?

    Shuttleworth: Yeah, but I don't think that's as definitive as that. If I look at the work that has been done in the KDE community, it is very vibrant. I use KDE on my desktop, I enjoy seeing the pace of change there, there is a lot of innovation in KDE4. I think the KDE guys have a point when they say their approach has made it easier for them to make leaps forward than the GNOME approach which has very predictable release schedules. The flipside to that is that this predictability and also the choice of the LGPL has made GNOME very good for business.

    Rather than saying: "GNOME wins, KDE looses" I'd like us to say: "How can we get this communities to sit down and talk to each other"? We really need to have both, stable release cycles and the ability to evolve quickly and make big leaps like KDE4.

    I'm very interested in finding out, how to get those two communities working closer together, how to get more collaboration, more sharing. Both at the level of technology but also at the level of best practices / processes. Talking to Nokia, it seems like they are interested in pushing this as well.

    Shuttleworth: Well they just acquired Trolltech, so it makes sense for them. A lot is going to depend on what Nokia is going to do from a licensing point of view. And separately what GNOME is going to do if Nokia makes the Qt-licenses effectively compatible with the GNOME vision, can they embrace Qt as a platform? So you would favor GNOME to switch over to Qt?

    Shuttleworth: Well, I think it would be perfectly possible to deliver the values of GNOME on top of Qt. There are licensing issues, GNOME is very much built on the LGPL, allowing companies to build their own products on a free software system, giving them some freedom and flexibility in their choice of licensing. That's very frankly been a huge drive for the adoption of GNOME by corporate ISVs.

    Whether we'll be able to have the FSF excited about something, have GNOME excited about something, have Nokia excited about something which makes life better for developers - that's gonna be the interesting challenge for me. I'd like to see both desktops focusing on a common infrastructure. And we've already seen that, a lot of the Freedesktop initiatives have been embraced by both projects - HAL, d-bus for instance.

    This also applies to other software projects, if you name your project g-something or k-something your are articulating a very specific user experience. Projects should really look to the whole Linux desktop and see how they can appeal to both sides. Recently there was a discussion about "decadence" in the GNOME community, discussing if GNOME is solely in a state of maintenance anymore. Do you think there have to be some bigger changes to get GNOME innovating again?

    Shuttleworth: I think GNOME really set the pace about good guidance, good release management and good stability for downstream developers. And that's very valuable, that's one of the reasons why we picked GNOME as the first desktop supported in the Ubuntu platform, that's probably also the reason why the majority of companies that develop for Linux use GNOME. But it's equally important to have a very clearly articulated strategy for how to we will introduce waves of innovation. And I think the KDE-guys have a point when they say, if all you do is have an everlasting commitment to a stable API/ABI and do releases once every six months, you can never make big shifts of innovation.

    It's not just as easy as saying we'll have GTK+ 2 then GTK+ 3, you really have to plan on how to introduce change into the platform. And I'd like to see more discussion in GNOME about that.

    We need to think of that as "what's best practice, how do I manage change?" All the things we criticize about Microsoft and Apple, they really try to solve real problems. And you can't run an old Windows application on a recent Windows version. You used KDE as an example for big leaps, but isn't there also a danger in that, seeing how the KDE-project seems to struggle with some of the unfortunate side-effects of such a very big leap?

    Shuttleworth: Yeah, that's a really interesting point. That's why I think it's not as simple as saying "six months release and than a big release which is perfect and then has no changes for two years". Because it won't be perfect the first time around as those big dislocations do hurt. That's why I think a very careful conversation between the lead thinkers, focused on real experience, focused on the real commitment to continue to deliver for the industry that stability, that predictability but also let's you introduce change. How you introduce change is very important. I think we should explore for example the idea of making sure when GTK+ 3 is introduced you can run in parallel with GTK+ 2, potentially even inside the same application. And that we don't say if GTK +3 is released it will be API/ABI-stable forever, cause it won't be perfect. So we might need to say: Lets GTK+ 3 iterate for a year or two and then make the API/ABI-commitment and drop the commitment on GTK+ 2. If you look on the desktop market today there is one operating system that is growing significantly and it's not Linux. It's OS X. What do you think is the reason for that?

    Shuttleworth: First of all, we should really understand this, as it's an important observation: The fact that OS X is growing, tells us that Windows is weakening. The fact that OS X is growing and Linux isn't, tells you that OS X is offering things that Linux is not. One of those is the pace of change, the level of innovation. You really have to give credit to Apple for driving innovation. Another of those things is their focus on the web as an experience. They recognize very strongly that the web is the killer application of the PC today and not Microsoft today.

    There is a real opportunity for us to deliver a great web experience, but we have to focus very strongly on getting this done. So OS X is more interesting for you than Windows?

    Shuttleworth: For me OS X is more interesting. I believe that free software is the most amazing platform for innovation, but I believe that that innovation also tends to follow a "lazy path", people often choose the path of least resistance, they want to express their ideas and they want to find the easiest way to do that. And at the moment we don't offer a particular easy place to go and express your technology.

    There's some exceptions to that obviously, if you look at Firefox plugins for example. Once Firefox got to the same level of functionality as Internet Explorer we saw an explosion in the number of plugins. That was driven by people saying "I have an idea on how to make the browser better". And what's the easiest way to achieve that? Go write a Firefox plugin! And I guess that should tell us a lot.


  What do you see as the main obstacles holding back the success of the Linux desktop?

    Shuttleworth: I think we don't yet deliver a good enough user experience. I think we deliver a user experience for people that have a reason to want to be on the Linux platform, either because of price or because of freedom. If that was your primary reason, Linux is the right answer.

    But if you are somebody who is not too concerned about price, who is not too concerned about freedom, I don't think we can say the Linux desktop offers the very best experience. And that's something we have to change, that's something I'm committed to work on, focusing increasing amounts of resources of Canonical on figuring out on how we actually move the desktop experience forward to compete with Mac OS X. Any concrete plans to share?

    Shuttleworth: I'm reluctant to say "This is what we are going to do". GUADEC is important to see, where we should help, where additional developer resources might make a difference, maybe it's not even about developer resources, we are trying to understand while we are here what would move the desktop forward.

    Second thing is, we just can't do this only on GNOME or just on KDE, we need to figure out on how to move the whole Linux desktop platform forward. I suspect if we hire a bunch of upstream developers they will be across both GNOME and KDE. A few months ago you started Ubuntustorm as a kind of Brainstorming page where users can provide their ideas and vote for the best ones. So how does this work from your side?

    Shuttleworth: We have a couple of hundreds of contributors to Ubuntu and most of them are motivated by their personal interests. To a certain extent, if something is hugely popular that argues in favor of contributing resources to. For example the media work we have done in Ubuntu, in making it easy to get additional codecs, that was a direct result of careful examination of the distro team of what people are asking for. Ubuntustorm is something that really drives this thinking.

    But it is still much more powerful if you show up with code or if you can convince other groups of developers to participate. One thing that we are trying to make sure is that the distro isn't a blocker to innovation. So that's why we have PPAs, where developers can have their own repository of packages which can be either modified versions of existing Ubuntu packages or completely new packages. The idea is: If you have an idea - go ahead and do it. You don't have to ask for permission, you don't have to wait. Create the package, publish the package, get users for that package. Still not everyone is able to contribute in that form, so are you continuously checking that list for interesting stuff?

    Shuttleworth: At the very top of the list - yes. At the Ubuntu Summit in Prague for instance their was a very significant look through the items on Ubuntustorm.

    But there is also a lot of amazing stuff that is not on that list and coming from the community, like Wubi, where you can install Ubuntu inside Windows. That would have never been something we would have conceived at Canonical, but now it's in there and it's very high-quality work. And it's made a huge impact on the community. One of the most popular items on Ubuntustorm is the one asking for a new default look. Originally this was planned for Hardy, why didn't this happen?

    Shuttleworth: Well, I think the LTS-release should be at the end point of the development cycle,so it's not a good point to change the look. Recently we have been testing a dark theme, it won't be the default in the final release, but we set it as a default to get feedback. And we got a lot of feedback, people saying "Oh my god, this is going to be the default and it is dark!". So dark themes are very useful for specific applications and people who focus on certain types of applications, content-oriented applications.

    I would like to see a fresh look, I'd like to see that done in partnership with the community, but I haven't yet figured out, how to do art in a community process. I think art requires strong leadership, to get free software art right, we'd have to have both strong leadership and strong community participation. And I haven't seen that happen yet. Linux distributions do tend to brand their software very heavily, from splash screens to specific icons. Wouldn't it be better to stick to the default GNOME-Look as far as possible instead of redoing everything in the own distro colors?

    Shuttleworth: I think in a very real sense "pretty" is the future, you want to share something with your friends which expresses a certain kind of experience. And art and design are very strong parts of that experience. I'd like us to do better in Ubuntu but also in free software in general. I think the Apple guys have a very good point when they say we should let designers lead the definition of the user experience. In the past you announced a strong partnership with Sun,especially that Ubuntu is going to be certified by Sun, this didn't happen, so what happened to that partnership?

    Shuttleworth: Well we didn't get certified, that's true. We did a port to SPARC, so we supported that for the Dapper-cycle. We have expanded the relationship, we made sure Ubuntu works well on their hardware, x86 has become more important. I think very basically if you are going to run Linux on SPARC you are doing this cause you feel you can support yourself. And if you want support you just choose Solaris and get support from Sun. So it wasn't really a big economical opportunity for us, but it was still very good to do and a valuable step forward for our relationship with Sun. But now it's x86 which is our focus in the relationship with Sun.

    The other big focus is around the software stack, around Java, which Sun has really been moving forward with their commitment to open sourcing. So in Intrepid for instance we may well see Java in main with the Sun OpenJDK and we work to have that certified so that it is a real Java-experience but it is also a real Ubuntu experience. So I guess our relationship with Sun has maybe become less visible, but it's got much broader. Thank you for taking the time to do this interview.

    (Andreas Proschofsky,, 14.07.08)

    Share if you care.