Where to run away from Google?

As many other open source believers, I was naive enough to believe that Google was somewhat better than other software giants. They have a quite loose relationship to open source. Many of their products are closed source and many have published source code, but are not really open (Android,…). But I believed that at least open standards and interoperability were important things to them. But after announcing that they’re going to kill XMPP support in their instant messaging service and especially removing federation (ability to communicate with other Jabber servers), they’ve lost my trust. And because they seem to back off from support of other open standards, too (changes in LDAP,…), I’m looking for alternatives to replace Google services I’m currently using.

What’s the most urgent is Jabber because Google will shut it down soon. I’m currently using Google Apps on my domain which is really good because my address is not tied to Google (or any other provider). I can easily come back to my previous mail and Jabber provider who offers Jabber on custom domains, too. It’s just matter of changing DNS records. What’s nice about it is that I can still use my domain for Google accounts since Google Hangouts are now closed to other Jabber servers and DNS should not be needed for routing messages to others users within Hangouts.

When I was thinking about possible Jabber solutions, I came to a conclusion that it’d be really nice to have a Fedora Jabber server and provide our users with an open IM service which is going to stay open. All FAS accounts already get fedoraproject.org mail addresses. Why not to have the same Jabber accounts, too? Having support for FAS in online accounts which would add the Jabber account to Empathy sounds like a great service. It could also be a unified platform for IM among Fedora contributors. I know we have IRC and Freenode, but it is not the same.

Anyway, I’ve once again learned an important lesson:

It’s very convenient to have tightly integrated services, but vendor lock-in is a damn bad thing. 

Spring Season 2013

I haven’t written a blogpost for a while. It was mainly because I was too busy with all the events I organized or attended in the last three months. So here is a little recap:

• Feb 23-24 – DevConf.cz 2013 - this is an event that took weeks of my life. I was the head of the organizing team and we went really large this year. There were almost 100 talks, workshops and sessions. We counted around 700 attendees and we haven’t had any significant problems, so success.
• March 2-3 – InstallFest 2013 – this is a traditional Linux event at Strahov campus of CVUT. Strahov has always played an important role in the Czech Internet and it’s called Silicon Hill. It has a strong Linux community. I delivered a talk on where Fedora is heading.
• April 10 – Afternoon with Red Hat in Bratislava – a set of talks introducing Red Hat and its open source projects and technologies to university students. We talked on Fedora, ABRT, openJDK, and MRG.
• April 15 – Afternoon with Red Hat in Prague – the same event as in Bratislava, just talks were different: Fedora QA, Ceylon language, and QML.
• April 17 – Red Hat Open House – another “day of open doors” in Brno offices of Red Hat. There were a lot of talks, programming contests, we held a F19 power management test day whose room was full all the time. I delivered two talks on RHT programs for students, community activities etc.
• April 22 – Afternoon with Red Hat in Ceske Budejovice – it was our first time in this city and we were surprised how many students came and how interested they were. We talked on Fedora, Fedora on ARM, JBoss, OpenShift, and Modern Linux Desktop. It was probably the only Czech university which has a lecture room with RHEL (not CentOS).
• April 23 – Presentation of Red Hat Thesis Topics at FIT BUT – we prepared another set of thesis topics for the next school year. Students can work on open source projects with us. At this event, we showed student what they could work on and tried to answer all their questions. BTW we have a new thesis management system, check it out
• April 24 – Presentation of Red Hat Thesis Topics at FI MUNI – the same event, just different Brno university.
• April 25 – Red Hat Presentation in Bratislava – another event in Bratislava, in fact just one building away, a different faculty. I talked on RHT programs for students and community activities.
• May 7 – Day of Industrial Partners at FI MUNI – career fair kind of event, we had a short presentation of Red Hat and then we were answering students’ questions about Red Hat.

And it’s not the end. On Monday, I’m going to EurOpen to talk on the transition from GNOME 2 to GNOME 3. And on May 21st, I’m going to LinuxTag 2013, probably the biggest Linux event in Europe. Life never stops

Fedora: Giving Up Product?

There is an interesting discussion going on in the Fedora Board and it gathers a lot of ideas. Some of them also say that we should give up defaults, or Fedora as an end product. I opposes such a direction and here is why:

Giving Up Defaults

Giving up defaults means giving up Linux newbies because it’d lead to the situation I call “new restaurant experience”. You go to a restaurant you’ve never been to and they give you an endless menu with tens of items usually strangely named. All you know is that you want a good meal, but you’re lost because you have no experience with the cuisine, you know almost nothing about the meals (except for ingredients) and you still need to choose something. Then the waiter comes to your rescue: “What meat do you like? Beef? Great, we’ve got this great meal with beef. You’ll love it! Would you like to give it a try?”, “Sure I would!” Or he could just say: “Beef? Great, we’ve got a huge selection of meals with beef, here see the section Beef.”  Would it help you? I can say it wouldn’t help me and when I’m in an unfamiliar location, I’m looking for restaurants that have simpler menus and predictable meals just to avoid such situations.

It works the same way with software. When my friend gave me a CD with Knoppix, I saw that Linux was quite nice on the desktop and I decided to give it a try. Knoppix was just a live distro, so I was looking for some more solid distribution. All I knew was that I wanted Linux for desktop. Someone told me that Mandrake was the best option for desktop and I went for it. I was glad that they had defaults (environment, apps,…) because I could not possibly make a qualified decision since I knew very little about Linux, and I trusted Mandrake that they chose a good selection for me. Mandrake’s default environment was KDE and I was satisfied with it enough to stick with Linux. After some time, when I was settled, I explored other options and found GNOME a better option for me. But that doesn’t mean I didn’t appreciate KDE as the default option at the beginning. It helped me.

Having defaults is about guiding. You tell newbies what you think is the best experience for them and it’s usually all they want to hear. Once they get more familiar with the distro, they can explore other options and find out that there is a whole world out there. Exposing the whole selection to new, unexperienced users is not helpful, it’s discouraging. The other day, one friend of mine told me that he needed Ubuntu or Debian to install one product that is supported only on these distributions. Because he had no experience with Linux, he asked which one. Well, I told him Ubuntu because I knew that was the quickest and easiest way to his goal: having that product up and running.  Just compare ubuntu.com and debian.org. Ubuntu gives you a very easy way to download and install it while Debian reveals all the complexity right at the beginning. Great for those who know exactly what they want, otherwise simply discouraging. And Debian still has defaults.

Having defaults is about focus. If you want to make a good product, you need to focus. It’s another thing Ubuntu did right (not any more with all that tablet/TV/mobile craze). It’s better to have one solid and working solution than ten unfinished and broken ones. If you have defaults, you know what really needs to work and you can focus on that.

Having defaults is about responsibility. A distribution is a huge selection of software. Something works better, something works worse. But it’s our responsibility that what we push to users as defaults is well maintained and has some future. I’m not sure if we can tell that about all desktop environments and window managers we’d have to equally offer if we had no defaults.

I believe having defaults is very important for Fedora Project. If we should have some default selection, it should be by use cases. You want a Linux for your desktop? Here is our product for desktop. You want to run Linux in the cloud? Here is our product for cloud. I know that choosing defaults is difficult and brings long discussions. But giving it up just because it’s difficult is like hiding head in sand.
Don’t get me wrong. I appreciate every new desktop environment, window manager, or application that is available in Fedora repos because freedom of choice is great, but having defaults doesn’t limit this freedom.

Giving Up Product

Making Fedora just a platform for other end products goes actually far beyond giving up defaults. Fedora would lose a lot. If you don’t have your own end product, you pretty much lose a lot of your visibility and brand. “Selling” a platform to users doesn’t make any sense because users (and most developers, too) don’t care about the platform what’s behind the product. They would use e.g. GNOME OS and just a few of them would know that there is actually some Fedora behind it and even fewer of them would care. Would it help bring more contributors? I don’t know, but I guess it probably wouldn’t. People get more likely attached to the product they’re using. While I like GNOME and I’m also a GNOME Foundation member, I’d rather switch to a different environment and stay with Fedora than stay with GNOME and switch to another distribution. This kind of attachment is very important for getting people involved and contribute. Without being the product people are using, we’d lose the ability to build such an attachment.

There was actually an attempt to build just a platform upon which others can build their products – Unity Linux. And it never took off. They never attracted enough developers while Mageia, another derivative of Mandriva which is also an end product, is doing much better. I still think a distribution like Fedora is the best wrapping for what’s called a Linux system. While e.g. GNOME is the face of the system, it’s Fedora who has the expertize from the kernel up to the desktop.

Another question is if any community would be interested in building a product based on Fedora. Why wouldn’t they choose Debian at the first place? By becoming just a platform, Fedora would lose a lot, but would we get something back, someone else on board? I doubt. And OS products generated from our own community? Regarding desktops, the GNOME part of the Fedora community might able to produce a solid desktop product, maybe KDE, too. But that’s pretty much it. I don’t see any other spins that are strong enough to build and promote products on their own.

Again, don’t get me wrong. I’d love to see Fedora as a great platform to build on, but I’d rather have Fedora as a great product to use and I don’t think that building a great product prevents us from being a good platform to build on. However, I’d encourage people to build things in Fedora rather than on Fedora.

And what would be my vision for Fedora?

A truly free and community general-purpose operating system that aims at people who create things and build solutions. It doesn’t matter whether they are designers, developers, admins etc.

Best Talks at DevConf.cz

There were almost 100 talks, labs, and workshops at Developer Conference 2013 in Brno. After the conference, we organizer a contest for best talks where attendees could vote for 3 favorite talks. And the results are…

1. Bryn M. Reeves – Who moved my /usr?? – Staying sane in a changing world
2. Lennart Poettering & Kai Sievers – What are we breaking now?
3. Daniel J. Walsh – SECure Linux Application Container
4. Lennart Poettering – The systemd Journal
5. Leslie Hawthorn - Negotiation theory for open source hackers
6. Lukáš Czerner – Local file systems update
7. Václav Pech – Pick the low-hanging concurrency fruit
8. Koen Aers – Raise your Java EE 6 productivity bar with JBoss Forge
9. Lukáš Zapletal – Java loves Ruby: Katello on TorqueBox
10. Jiří Olša – perf profiling

Congratulations to winners! We’re currently looking for the best way to award them because they’re from all over the world.

BTW almost all talks are available on Youtube. The most popular talks based on Youtube views are:

1. Lennart Poettering & Kai Sievers – What are we breaking now? – 1345 views
2. Lennart Poettering – The systemd Journal – 714 views
3. Vratislav Podzimek – The technology beyond Anaconda NewUI and 3rd party extensions – 336 views
4. Aleš Kozumplík – Hawkey and DNF: the next-gen Fedora Packaging tools – 292 views
5. Tom Callaway – Improving the Fedora User Experience with Design Driven Methodology – 274 views

The most attended talk at the conference was “Lennart Poettering & Kai Sievers – What are we breaking now?“. The attendance in a room for 200 people was waaay over its capacity.

That was some popularity stats. Hopefully, we’ll have even more interesting talks who you’ll enjoy at Developer Conference 2014.

FAm Budget and Governance Reform

We’ve entered the new fiscal year in which we’re facing a lot of changes and new things. No, I’m not talking about GNOME 4 or systemd 2 I mean changes in governance of Fedora Ambassadors. In FAmSCo, we worked on one of the biggest changes in how Fedora Ambassadors (and other contributors, too, because they’re not excluded) can spend money. But before I start explaining all the changes, let me state problems we were facing and reasons why we did the changes:

• Our spending was decreasing. It sounds like something good, but in fact, it meant that we were doing less and less activities. Financial deporting wasn’t good (it was something we couldn’t really influence) and the whole community including FAmSCo lived in uncertainty how much money we have spent and how much we still can spend. So we rather didn’t spend. And because the classic corporate rule says “if you don’t spend it, you’ll lose it”, Fedora’s budget was shrinking.
• FAmSCo evaluated almost all funding requests, so instead of focusing on “big” problems, we were dealing with day-to-day problems which is something a steering committee is not for.
• Decision-making was pretty centralized and local communities were not so involved.

And here are the changes:

• BUDGETING – first we wanted to remove the uncertainty which caused that we were afraid to spend money and do activities. Before that, FAmSCo was just approving requests and kept asking CommArch/OSAS if we still have money left and mostly we didn’t get any answer because of slow financial reporting. Now, we actually do budget planning and ask OSAS for an annual budget for the upcoming year. It has two positive outcomes: 1. we actually know how much we can spend and if we properly track our expenses we don’t even have to rely on Red Hat’s financial reporting and know how much money we have left. 2. if you’re making a budget and asking for money, you have to have an idea how to spend them. So instead of planning just expenses you actually plan activities on which you spend money. Of course, you can’t plan everything. There are always unexpected activities, new opportunities. We don’t want to turn them down just because we don’t have them in the budget. That’s why there is a treasurer/budget wrangler in each region that adjust and balance the budget. Budgeting is new to us, so this fiscal year is going to be a lot about gaining experience.
• POWER TO REGIONS! we also believe that the power should be closer to community members. So we decided that the overall budget would be divided into 4 regional budgets: NA, EMEA, LATAM APAC. Each region had to think about what activities they’d do this fiscal year and how to fund them. It really helped some regions because until then they had hardly done any planning and had problems to spend money allocated to them. And I must say it helped all regions in the end. Now, regions are responsible for their budgets and they also evaluate and approve funding requests that belong to the region.
• NEW APPROVAL PROCESS – because we gave regional communities the right to approve requests, we had to change the approval process. It was really difficult to find rules that would be applicable in all regions. We really didn’t want to damage what was working well for regions. So we decided to give them freedom within some boundaries and provide referenced rules. The boundary is $2000 limit per request. If it’s higher it needs to be approved by FAmSCo. The referenced rules are that expenses up to $500 can be approved on peer-review level (by a FAmSCo member, or community credit card holder), and expenses between $500 and $2000 needs to be approved by the local community (votes from 5 ambassadors at a regional meeting). Those limits and rules are not enforced by FAmSCo, regions can set whatever they want as long as they ask FAmSCo about requests over $2000, stay within their budget, and do everything transparently.

On the picture above, you can see the overall process. First step is budget planning. Regional community members have to get together and plan what they’d like to do in the next year and how much money they will need for it. Once their plan is approved by Fedora budget owner, they have a budget and know what events/activities/… they have money for, e.g. $2000 for FOSDEM. The approval process is about decisions how exactly they’d like to spend the money. How much do we need for swag at the booth at FOSDEM, is this contributor worth sponsoring airfare to the conference? All those requests should be in separate tickets and approved separately. The third step is payment itself. If it’s reimbursement it’s where community credit card holders come to the stage. They should make sure that all conditions under which the expense was approved were met and make the payment happen.

FAQ

I’m a community credit card holder, trusted by Red Hat to make payments. Why do I have to ask for an approval?

You surely are trusted by Red Hat to make payments, but payments are not expenses. Payments are done from a bank credit, expenses are done from a budget. Every payment has to match a budget expense in the end. You can find different budgets to make expenses from, but if your payment should end up as an expense in one of the regional budgets, in other words, become their expense, it’s fair to play according to their rules.

FAmSCo is taking power from us!

In fact, FAmSCo has given up a lot of power. We no longer approve most requests. FAmSCo now mainly serves as a supervisor making sure the whole system works and no one abuses it. Most of the power is now in the regional communities, much closer to community members.

Why should FAmSCo dictate us how to spend Fedora Project’s money? It’s not the emperor of Fedora Project, it only takes care of the Fedora Ambassadors program.

The whole governance and budget system I’m talking about applies to the regional support budget which has always been handled by ambassadors. But that doesn’t mean other contributors are excluded. They have an equal right to ask for funding from this budget as long as they follow the rules.

Why do we need rules anyway? We’re a community of Fedora enthusiasts. Good intentions should be enough.

Well, you’re a community of hundreds of people, maybe thousands. Such big communities can’t work without rules. That’s a fact. The governance and budget system helps us plan, prioritize with limited resources and make sure money is spent wisely. All our expenses also end up in RH accounting which has to follow many rules and legal restrictions. We tried to design the system to be as least bureaucracy as possible and we’re open to improvements if you think that there is too much red tape in some processes.

Useful links:

Reimbursements

EMEA Budget for FY14
NA Budget for FY14
LATAM Budget for FY14
APAC Budget for FY14

Another distro popularity polls

Last summer, I published results of survey organized by the biggest Czech Linux portal where Fedora was the 3rd most popular Linux distribution. Today, the second biggest Czech Linux portal ABCLinuxu.cz published results of a very similar survey. Shares are slightly different, but Fedora was again the 3rd most popular distribution. The TOP10 is:

1. Ubuntu 37.09% (53)
2. Debian 18.78% (17)
3. Fedora 18.39% (12)
4. Linux Mint 14.03% (10)
5. Arch 13.36% (9.5)
6. (open)SUSE 11% (7.4)
7. Gentoo 8.68% (6.7)
8. Mageia 2.51% (<2)
9. Slackware 2.20% (<2)
10. Mandriva 1.18% (<2)

The number in brackets indicates shares from the survey done by the other portal in summer. Shares are slightly different, but order on the first seven positions is the same and there are the same distributions in TOP10. The other portal – root.cz – is read by broader audience while ABCLinuxu.cz has more die-hard Linux fans among their readers. You can see that Ubuntu has a much larger share outside the core Linux user base. On the other hand, at least in the Czech Republic, it has lost some core Linux user base because in 2010, Ubuntu had 44.3% in the same survey. Over 18% for Fedora is a good result. It’s definitely on the rise in the Czech Republic which is not such a big surprise because we promote Fedora quite a lot and the largest engineering office of Red Hat also has some impact.

What’s interesting is distro popularity among groups of people with different length of Linux experience:

As you can see, Ubuntu is very popular among Linux newbies. Almost 60 percent of them choose Ubuntu to be their first Linux experience. Quite surprisingly to me, Fedora is the second most popular distro among newbies. Fedora has pretty much the same popularity through all groups, but it’s especially popular among newbies and then among people with >15 years of Linux experience. I suppose those are mainly users that started with Red Hat Linux.

There were also other categories:

Server: Debian 50.92%, CentOS 23.61%, Ubuntu 16.22%, RHEL 13.79%, Gentoo 8.87%, (open)SUSE 7.24%, Fedora 5.55%, Arch 4.81%, Slackware 3.33%. So Fedora is much less popular on servers.

Enteprise: RHEL 75.68%, SUSE 22.77%, Oracle 4.94%, Mandriva 1.55%. RHEL is dominating this category.

The survey also showed that the Linux user base is getting older. 43% of people who participated in the survey are over 30 and only 25% of them are students.

FOSDEM, F18 release party,…

Busy times. Just a week after I returned from my adventurous journey to FUDCon Lawrence, I had to get ready for FOSDEM 2013. The night before leaving for Brussels, I found out that no one was bringing equipment for the booth and swag. So I stuffed my suitcase with tablecloth, stickers, buttons, and a hundred of baseball caps. Fortunately, Jarda Reznik was going to Brussels by car, so I gave him the heavy stuff – rollup banner, mugs, and DVDs. Even though our booth was decent in the end, the organization wasn’t very good this year. FOSDEM 2013 is #1 event in EMEA and we should give it more. Booth staffing didn’t work very well either. We always had some people there, but I found very unfortunate that people who didn’t get a penny from Fedora Project had to stay at the booth while some people, who got sponsored, spent there very little or no time. Next time, we have to make 100% clear that people who get sponsored go to FOSDEM to primarily help Fedora Project which pretty much means helping at the booth.

FOSDEM was crowded like always and we ran out of swag and DVDs at the beginning of the second day. Then Peter Robinson saved us and loaned us a brand-new model of OLPC with touchscreen. That was a new attraction of our booth. A lot of people came to our booth and asked if they could buy a Fedora T-shirt. It was really heart-breaking to turn them all down while all other projects were selling stuff like T-shirts, mugs, hoodies etc. I brought some Fedora mugs and baseball caps, but it was enough just for Fedora contributors I met. Inability to sell, and thus produce more expensive stuff such as T-shirts and hoodies is our big marketing disadvantage. Not only are we unable to satisfy our fans, but it also harm our marketing because people with Fedora cloths become walking advertisements.

Anyway, in spite of all those problems, my feelings from FOSDEM 2013 are generally positive. It’s a bit too crowded and large event to my taste, but where else can you meet so many people from the open source world? I was really glad I met several new Fedora contributors that I hadn’t met in person before.

The day after I returned from Brussels, we held a Fedora 18 release party in Brno office of Red Hat which I was organizing. It was not the ideal date because there is a break between semesters at universities and students have always the biggest group of attendees. But we were still able to attract several dozen people. This time, we had talks on GNOME 3.6 (by me), How to test Fedora (by Kamil Páral), DNF (by Aleš Kozumplík), PostgreSQL 9.2 (Honza Horák).

The empire of Fedora QA needs you, Kamil Páral says.

Another event is just in two months – Developer Conference 2013 which now takes most of my time because I’m one of the organizers.