Just after Christmas, Neowin was given the chance for a community Q&A with Paul Frields of Red Hat. Paul has worked with Fedora since 2003, and became the Fedora Project Leader in January last year. Along with a few other things, Paul is also chairman of the Fedora Project Board, which makes decisions on how Fedora will move forward.
As with the Q&A with Joe Brockmeier, Neowin"s forum members came up with a great selection of questions.
One of the things I have always noticed about Fedora is the default desktop wallpaper, which gives a real sense of effort and refinement when you see it. Did Fedora choose to go this route, or did it happen by chance? Is there any particular thought process or symbolism behind the images and how long does it take the artist to create them?
The artwork for Fedora 10 and many previous releases has been created by the Fedora Art team, through a collaborative, open process led by Mairin Duffy and involving many of our volunteer contributors. For Fedora 10, they ran a three-round process which started by soliciting theme concepts and drawings. Each contributor could post a sketch or design idea, which the community would analyze and critique, to help strengthen the design. In the second and third rounds the artists refine their work, after which the Art team comes to a consensus on which theme is deemed the best suited for the release. Then the design is used to create all the supporting artwork like the various splashes, banners, buttons, desktop backgrounds, and so forth. We also use it throughout our marketing materials for maximum impact.
It"s also important to note that all our artwork is created with 100% free content, and using only free software tools in Fedora like GIMP and Inkscape. So when people ask how good the artwork tools are in Fedora, we can point right to the artwork made by the Fedora Art team as examples.
In the Fedora 11 release, the process will still be done transparently, but a little differently. We have made the release naming process occur a little sooner after the release of Fedora 10, so that the Art team can use it to inform their designs. I"m not sure exactly how "Leonidas" will be interpreted, but they"ll probably tie it somehow to strength, or to Greek architecture as a pinnacle of design, or something like that. I"m as eager to find out as everyone else!
(Chicane-UK) It felt to me like 2008 was an excellent year with distributions like Fedora and Ubuntu gaining some real ground, and firmly finding their feet - especially now with a surge in Netbook sales, people are looking to (and finding) a great product in Linux. Do you see this continuing on into 2009 and beyond? Will Netbooks continue to be prime real estate for Linux vendors?
(Symod) We"ve seen that some major distros are taking into account the netbook market by releasing specialized versions of their distributions (like Ubuntu Netbook Remix). Considering that Google is doing a similar thing with their Android OS (which should soon premiere on a netbook as well), and that the portable device market is growing, what are Fedora team"s views on it? Any special version we might see in the future?
(kyro) How does Fedora look at this new trend of Netbooks?
I think netbooks have become a very interesting market for Linux. There"s clear consumer interest in Linux on the netbook platform, and as people probably know, Fedora is used as the basis for a couple of the netbook OS platforms out there, like Linpus. It will be interesting to see what Microsoft does with Windows 7 on this platform
to try to stifle that growth. Essentially though, people are not going to want to pay a big markup for Windows on what"s supposed to be a very cheap platform.
Two ideas foremost in my mind are (1) using the netbook platform to help us identify resource hogs on the desktop and slim them down, so the entire distribution benefits and upstream as well, and (2) seeing if the One Laptop Per Child machine"s Sugar desktop, or its next evolution, might be viable for these devices. I"m not sure the standard Linux desktop UI works well on netbooks with their limited screen size and resolution. It would be interesting to see how we can optimize that experience over time.
We provide a complete set of tools for creating Fedora Remixes, combinations of our software, with or without other add-ons. Any community can also use the new "Fedora Remix" mark to label their Remix, and combine Fedora with other software as needed to create a compelling product for a netbook or other hardware platform. You can find out more information at https://fedoraproject.org/wiki/Remix .
Part of our purpose in creating the remix capabilities in Fedora, which we"ve had for several years, was to enable people to take advantage of the fact that Fedora is, and always will be, only 100% free and legally redistributable software, no matter where you live. By not playing tricks with items that have questionable legal status in certain locales, we can enable you to change the platform into something that works for you and the communities in which you work and live. Now, with the Fedora Remix logo, you can create that special remix, and say, "I made this new product using Fedora," and point people back to the Fedora Project to help work on the bits underneath. So it"s really quite easy to create, say, the "Netbook 2000NBX Fedora Remix," specifically for that platform.
(Chicane-UK) Many people (rightly or wrongly) attribute the increased success of MacOS and a number Linux distributions to Windows Vista. Microsoft look to be trying really hard to make sure Windows 7 doesn"t launch into the same kind of hate-storm. What are vendors such as Red Hat doing to make sure that they"re ready to compete with Windows 7? Are there any killer features on the horizon (beyond the good stuff we"re already used to)?
I don"t create product strategies for Red Hat, but I do get to lead the project that is the upstream for Red Hat Enterprise Linux. So if you"re using Fedora now, you"re seeing some of what"s going to eventually end up in some form in Red Hat Enterprise Linux 6. The way this works is that features in Fedora at a certain point are branched off for RHEL. Hundreds of Red Hat engineers and QA people then beat on these bits, and make sure that they are appropriately poked, prodded, and tested to be ready for enterprise-class customers to consume. That"s a huge amount of work and it"s why RHEL only comes out every couple of years.
In my mind, the incredibly compelling virtualization platform we created back during the RHEL 5 timeframe has improved by leaps and bounds with KVM, the Kernel Virtual Machine. The fact that every single Linux kernel now carries with it the ability to act as a complete hypervisor basis, without substantial added maintenance, is a huge win for enterprise customers. We"re not kidding when we say that if you look at Fedora, you"re looking at the most relevant technologies for the future of open source. Our contributors created libvirt and virt-manager to make it easy for people to use the older Xen virtualization, which is very costly from a code maintenance perspective, the same way they use the newer, slicker KVM. In the future I really see KVM taking over as the virtualization platform of choice. And because we have all the migration tools needed to go from one to the other, thanks to the benefits of open source, it"s a completely painless transition for everyone.
On the subject of Windows, the Windows architecture has been notoriously bad for things like virtualization, because there are so many enablers out there for violating strict rules on how the hardware gets touched by drivers. People are used to a certain amount of flexibility in Windows that is actually really bad for the stability you need as a virtualization platform. And the more that Microsoft cuts into that flexibility so they can promote the stability side, the more they anger the customer base they"ve built up.
(Fish/kyro) How much actual influence on Fedora does Red Hat have, and vice-versa - how does Fedora relate to the current Red Hat desktop OS? Has there been any package added to release on behest of Red Hat?
We have thousands of contributors in Fedora -- only some of them are Red Hat employees. The Fedora Project is constructed in a way that allows anyone to create new and compelling features in the distribution, not just Red Hat software engineers. No one, including Red Hat, has to sneak in any package to the release. All the work is done in the open, with all commits and packaging work available on public feeds like mailing lists and Trac tickets.
This is an important differentiation between Fedora and some other distribution projects. We do all this work in the open source community, and Fedora only uses 100% free and open source software and processes in everything we do. Watching the mailing lists and the other various data feeds for code and packaging will tell you everything that"s happening in Fedora, which makes life much easier for everyone.
As far as governance, the majority of the few steering committees and policy making groups are 100% elected by the Fedora community at large. That means that anyone can get involved in driving Fedora processes. And because we re-elect those people at strictly regular intervals, it makes them very accountable to the Fedora community.
The Fedora Board, the top of the governance food chain in Fedora, has 9 members, and 4 of them are appointed by Red Hat, and 5 of them are elected by our community. Those numbers are not as meaningful as you might think, though, because Red Hat currently has community volunteers in 2 of the appointed seats, and there are Red Hat employees who ran freely for elected seats, and the community voted for them. So we have a good mix of people on the Board whose most important quality is that they care a lot about the Fedora community -- regardless of what comes after the "@" sign in their email.
Our community appreciates that care and time that people put into Fedora, no matter whether they work for Red Hat or put in volunteer time on Fedora; anyone can be involved to whatever extent they prefer. This model has promoted a great sense of community in Fedora for a long time -- and other distribution projects have used our model to inform their governance. The openSUSE community went to a similar model recently, citing Fedora as an example of how to do this the right way. So I"m proud of that fact and we have no problem with people using and building on these ideas, just like with our software.
(Vieira) How is the Fedora Team working to make the project unique? What features will we find in Fedora"s next release that we won"t find in other distributions?
What makes us unique, as I mentioned earlier, is that we do everything the open source way. You"ll never hear us announcing that Fedora is open sourcing a product we use, because we *start* with 100% free and open source, always. I think building a free software distribution without using free software completely defeats the point of what you"re doing. Anyone should be free at any time to look at the Fedora Project, say, "I can do that better," and then fork whatever they want to do a better job. Our job, as Fedora, is to make sure they never need to do that.
We do all our work in upstream communities so that work can improve free software functionality for users of all distributions, not just Fedora. That"s a very important part of working in an open source fashion. It"s why you"ll find features like NetworkManager, D-Bus, HAL, PolicyKit, and others in many major distributions. If you want to see a whole list of things that originate in Fedora, check out this page on our wiki: https://fedoraproject.org/wiki/Red_Hat_contributions
Sometimes people are so eager to make their project seem better that they worry about aggregating more than they worry about originating. Taking copies of upstream projects and then patching them, sometimes sub-optimally or in ways that subtly break other pieces of the system, rather than working with the upstream for improvement, goes against the open source methodology and I hope that Linux distribution projects as a whole can outgrow that way of thinking. Architecture and programming decisions should be made with the upstream through open communication and finding the best ideas, not through surprise announcements. I would say that if you found a lot of features in Fedora that aren"t found elsewhere, that would represent our failure to work with software communities in the true open source fashion.
(Vieira) What is Fedora"s top priority for 2009? What goal do you want to achieve the most?
I would like to see a stronger quality engineering effort built around Fedora. We had a couple instances in the last six months where end users had a broken package update experience for a few days at a time. We were able to repair that easily, but it shouldn"t ever have happened in my opinion, and I"ve been talking with different people in Fedora about how we can improve on how we deliver bits to our users. The good news is that we are developing some better automated tools to help prevent a recurrence, and many of the Fedora community people are talking about quality as a concrete goal for Fedora 11 and 12. To do that properly, we need to answer basic questions like what quality is in a quick release cycle, how we know when we have it, and what to do if we don"t.
(Vieira) Would the Fedora team consider changing the release method to a rolling release instead of the actual six month scheme? If not, why not?
That"s an interesting question. Our six-month release has been a major reason why we"re able to put out compelling features in a "best of what works today" fashion with each release. Taking a longer release cycle didn"t tend to make for better releases; it just meant there was more slack time, and the release looked about the same in the end. A rolling release, though, has an opposite problem -- it vastly multiplies the possible end-user experiences in a way that is more suited for only hardcore developers.
While I think there is definitely a place in the Linux ecosystem for distributions like Gentoo, our users tend to be enthusiasts who still want to spend more time on their system working on something other than the system itself. For people who want the latest versions of everything, and don"t care as much about whether they system is a stable release, there"s Rawhide. You can usually install a system from our Rawhide repositories and watch as features and code changes during the release cycle. From Rawhide we draw the distribution in its test releases (Alpha, Beta, and Preview) on its way to the final release, like the upcoming Fedora 11. Our Fedora 11 Alpha, by the way, is due for public release on February 3.
(Vieira) There are people that consider that Linux"s main desktop environments (Gnome and KDE) try to mimic the behaviour, style and features of Windows and OS X. What do you think about this?
I actually installed Windows in a virtual machine a few days ago for a specific test someone had asked me about. I found that it was a very different experience from what I have in GNOME on Fedora 10. In fact, it was often incredibly confusing, with too many links and choices cluttering up the interface. I"m not a MacOS user and can"t speak to that specifically, but I think there"s a lot of thought that goes into Linux interfaces that is not simply aping prior designs. I"ve listened in on developers with the Red Hat Desktop team, for example, and designers working on better interfaces, enough to know that they"re not trying to imitate but to surpass.
(kyro) Can you name any package or service in Fedora which hasn"t got that publicity among users but is very useful for desktop users according to you? Any personal favorites?
I think that GNOME"s Dasher utility, for use in extreme accessibility situations, is really neat. In a case where a user may have highly limited (or zero) mobility, or in an interface with a single directional input, this kind of application is very important. It shows how Linux is really reaching out to embrace people and situations beyond the simplest cases.
I also love Zim for keeping desktop notes organized in a wiki-like way. There"s an application called Publican for taking DocBook XML documents and turning it into beautiful HTML and PDF versions, by building on top of other available tools. It takes some learning but it"s very compelling once you do that. And finally, I think people need to know about Preupgrade, an application that helps you update your Fedora system from one release to another. It downloads packages in the background, so you can keep working in the meantime, and when you next reboot, the system finishes the upgrade for you. There"s more information on our wiki at: https://fedoraproject.org/wiki/Category:PreUpgrade
(Vieira) Emacs or Vi? Gnome or KDE? :P
GNOME, and Emacs (although I know some vi and also use gedit). I use Mutt for email along with offlineimap, and although I like Firefox, I sometimes use elinks as well. I can operate in a text terminal pretty effectively without X, but I still prefer a GUI and don"t think of myself as a terminal elitist.
(kyro) There was a recent rumour that Red Hat is going to introduce Desktop OS (although I think they already have Desktop OS on their only shopping list), later it was announced that Red Hat will not release any new Desktop OS (Source:- Cnet) and simply doesn"t expect any viable business market for Linux Desktop OS. How did this news impact the Fedora team?
Haven"t you asked quite a lot of questions already? ;-) But seriously, this didn"t really impact the Fedora engineering team at all. There"s also a dedicated Red Hat Desktop team, and I can"t speak for them, but I can tell you that they have continued to pour their hearts and souls into dozens of new Desktop features regardless of what products Red Hat has been selling. Red Hat is absolutely committed to producing great desktop user capabilities, regardless of its product line. Again, refer to this page for information on features which Red Hat continues to contribute to the free software community through Fedora: https://fedoraproject.org/wiki/Red_Hat_contributions
(kyro) What are the top requests by Fedora users for the next version of Fedora?
We don"t generally do a survey. I think the one thing people tend to want is a longer release cycle, or a release with a longer support lifetime. But that"s exactly what is provided by RHEL, if you"re a business customer, or CentOS, if you"re a hobbyist with no need for a support contract beyong community help. Trying to produce both a constant stream of innovative features and at the same time a stable, unchanging release platform is a losing strategy. Red Hat used to do business this way, up until about 2002, and they called that product Red Hat Linux. Splitting the focus into a faster-paced, community-driven, feature-rich platform (Fedora), and a slow-paced, customer-driven, long-lifetime platform (Red Hat Enterprise Linux), is what made Red Hat a successful and profitable business. That success has allowed Red Hat to devote so many resources to directly developing code wiht the free software community.
(kyro) Ubuntu has lots of (comparatively) loud users who are new to Linux and are very excited seeing the prospect of it and the mileage they get from its use. Lots of developers are seen adopting Ubuntu as their default OS and they of course release the source code too, which can be rebuilt in Fedora. Have you personally seen any direct impact on Fedora and its release because of this adoption of Ubuntu by developers?
In the case of Upstart, we found there was an interesting idea and codebase being developed by someone at Canonical, although when he started the project, I think he was doing it in his spare time. We"re still working on pushing that feature into a next-generation replacement for init in Fedora, but it has some fascinating potential.
I think the community has pretty comfortably settled into an idea that Ubuntu is the beginners" Linux, and that when you"re a more dedicated developer, you move to something like Fedora or Debian. I don"t think that"s because Fedora is less user-friendly; I"ve run Ubuntu and Fedora (and openSUSE for that matter) and find that each platform has quirks and problems that its fans tend to either ignore, work around, or not notice. I think it"s because when people understand more about how free and open source software work, they would rather use a platform and participate in a community that has a very mature attitude toward contribution and supporting the upstream communities that provide, let"s face it, 90% or more of what"s in any distribution.
As I pointed out before, differentiating on the code base is actually a losing strategy. It"s less maintainable, promotes poor attitudes toward upstream developers, creates the appearance of open source being a mishmosh of functionality, and is ultimately not a sustainable way of doing business.
(kyro) How do you see the future of Linux adoption and Fedora"s role in it? Are there any strategic plans at Fedora to expand user-base? With Apple"s increasing user base, do you see Apple and Linux competing head-to-head in the future (Mac VS Linux ads?) or I am thinking far too ahead?
Yes, you"re thinking too far ahead. ;-) I think Fedora"s role in Linux is as the primary developer of relevant new technologies, and it"s why we tend to attract so many skilled develoeprs and enthusiasts to whom software freedom and innovation is so important. Apple"s user base is a very different group -- groups who are (1) not really that interested in technology other than as something that helps them with productivity, (2) have zero interest in freedom as a concept in software or communities of use, and (3) apparently have unlimited or at least large quantities of spending cash, given the incredible markup on Apple gear.
Point (1) keeps them from being as big a hit with technologists, and point (3) means they"re not going to be able to penetrate the BRIC (Brazil, India, and China) regions where free software is such a hit. As we spread the word about (2) that also means we have the opportunity to grow the market for Linux. Beyond these points though, I think there"s a place for Apple as a proprietary, design-centric technology vendor to co-exist with Linux, but it sure would be nice if we didn"t have to work so hard to fight their product lock-in.
(kyro finishes up the Q&A for us) Any message to Fedora and other Linux users? Can you tell us anything exciting to come in future Fedora releases?
Our release feature process, like *every* process in Fedora, is 100% open and anyone can see its progress at any time. Our upcoming features are being tracked here: https://fedoraproject.org/wiki/Releases/11/FeatureList
The features we"re working on for Fedora 11 include:
- 20 second startup
- Integrating PolicyKit with CUPS for a smoother user experience
- DeviceKit, to make hardware devices easier to use for desktop users
- DNS security, to provide integrity and authenticity of DNS data
- ext4 as the default file system in F11, for unparalleled scalability
- Fingerprint reader upgrades for secondary authentication with libfprint
- Multiseat, for two or more users to work on the same machine from separate keyboard/video/mouse units
- Presto, to add yum support for downloading only RPM changes rather than whole new packages
- Python 2.6, to prepare for later migration to Python 3000 (also known as Python 3)
- TightVNC, a faster, improved VNC client for remote system use
- Volume control, providing a better user experience for sound without all the endless hardware/software mixer twiddling
- Windows cross-compiler, allowing developers to build and test full-featured Windows programs without having to use Windows
- GNOME 2.26
- KDE 4.2
- Xfce 4.6
And of course there will be numerous feature upgrades and changes coming from the thousands of upstream software communities whose work is promoted in Fedora 11. Our release date is currently set for May 26, 2009, but you can help by testing the Fedora 11 Alpha that arrives on February 3, and filing bugs for problems you observe.
If you"re a Fedora user, it"s a very small step to become a Fedora contributor. Participation is how free software works, and you can help, by doing anything from filing a bug to writing a wiki page. Come to https://join.fedoraproject.org if you"d like to help free software continue on its path of world domination!
Thanks to Matthew for this chance to talk to the community.
Have a great week, everyone!
- Paul
-----------------------------
And thanks to Paul for taking the time to answer our (many) questions, as well as all the members who contributed to the Q&A!