 table with different representatives from different distributions and now I'll leave you with Andreas Stiehler. Thank you very much. Originally Makos tried to organize this round table but because he's not here and I feel somehow obliged to this round table I tried to take over this role and I hope I can manage it as he would done it probably. I want to like to introduce members of this round table. You see Martin F. Kraft. He will speak for VCSPKG.org. We had a talk seen yesterday. I think I don't have to introduce Mark. He was talking to you some minutes ago and then we have Cesar Gomez from the Linux people, Florian Meijer from the Munich distribution in Germany. It's called Linux. We have Holger Levson. He is speaking for Debian Edu and Kasten Debian distribution and Bide tries to fill out the role of Debian here in this. No, it's not really the end. I'm just extending the table a little bit. We are not really prepared for so many people. So I would like you to have some short introduction of three or four minutes for each of you. Do you want to start? So good morning. My name is Martin Kraft and I'm here to represent as Andreas already said VCSPKG.org which is a website that I put up in first half of this year. The goal of this website and I hope I don't bore everyone if you've been to my talk yesterday but let me just briefly give a one minute summary of what it is all about. The goal of the website is to find out how we can work better together with other distributions or how we can work on a cross-distro level. I'm actually looking only on the technical side of things so it's very much patch management oriented but because we are dealing with software and because we are dealing with packages and every one of us in the distros as well as in the derivatives are essentially doing the same task by maintaining packages there's a lot of potential for or there is a lot of overlap and there's a lot of potential for while making use of increased manpower. So I'm very very interested to cooperate with other distros and obviously our derivers are going to be the easiest ones to cooperate with simply because the differences between what we do and what our derivers do are necessarily small. It is after all still a Debian based system if it's derived from Debian. I would like to make one more point which is something that you Mark also mentioned earlier on in your talk and that is that we should optimize for collaboration and that we that in order to scale we need to be able to support others that are trying to use our work and build or I think you called it experiment with our work. I think if I look back at the last four years of Debian and Ubuntu the relationship then one of the things that strikes me is that even though Debian always felt that we're being ripped off that Ubuntu is bad that stuff's going wrong and that we're losing users and all that kind of stuff I think the reason sorry the reason why we haven't been able to cooperate are of course multi-fold but one of the big causes is that Debian has been expecting Ubuntu and by extension every single one of its derivers to do this work to do the optimization for collaboration for Debian so we weren't really able to or willing to step out and make it easy for our derivers to use and feedback use our archive and feedback into our archive and I hope that future discussions are going to prove fruitful so that we can support those that are building on our work so that's what I'm trying to do on a technical level and that's it for me. I will just pause the microphone right on because I think I've been introduced today already. Hello my name is Cesar Gomez and I'm proud to be here representing the Kenu Linux team of the Junta de Extremadura. Most of you know of us but for those who don't, Extremadura is a region of Spain with 1 million inhabitants and more than 10 years ago the regional government decided to start a new information society project. That project was meant to be a good project in terms of having to, they wanted to have one computer for every two students, they wanted to build a business incubator and they wanted to build 50 telecenters known as new knowledge centers there in Extremadura and so what we need was framework and a software that we could fully control and that is the reason we developed the new Linux version. It is of course based on Debian because we think that Debian has a strong community of good developers and because Debian doesn't rely on any company and that is important for us because we are not hiring the development of our own distro and we are developing it by our own so and I am also here because we are really, really interested in collaborating with other derivatives distros and we have been doing so for the last five years and especially during the last three years we have been organizing a lot of Debian work sessions there in Extremadura and maybe Holger could tell you about the advance we have been doing between Debian Edu and GNU Linux and so what we want to do is to continue collaborating and to keep the boom going forward. Thank you. Thank you. My name is Floremeia. I am working for the city of Munich. Many of you have probably heard about the project. I carried off its 14,000 desktops. It is like many office objects like forms, macros, which have to be reported in some way reported them to open office and much business operations, which we have to emulate, do things like we M-W-Y and Q-M-O and our motivations are quite similar to those of Extremadura people Linux so we are pretty interested in collaboration. I know personally that we haven't done that yet very much. We are opening patches, we are submitting patches, but I am personally as development lead missing the direct communication channel and so I was very interested in Martin's and Mark's talks, Martin's talk from the technical and Mark's talk from the social side of things so maybe we can get or I am pretty sure we can get some inspiration at this step conf and get things further and collaborate. Thanks. My name is Holger Lipsen. I work on Debian Edo. Debian Edo was started in 2001 as a project in France to customize Debian for education. At the same time people in Norway started Skola Linux, which had the same aim, so within the next two or three years the project, two projects became one, so now we have two names now and since last year the Linux people also merged, started merging their stuff into Debian Edo. The Linux people, the Linux is distributed in two forms. There is the distribution which they have for libraries and governmental use and for private use and the Linux which is used in school at the moment only the Linux in school is merging with Debian Edo. Debian Edo tries to work within Debian. We try to get all our changes in Debian and try to prepare or help developers learn the skills needed to maintain packages in Debian. So we are kind of a, we try to be a kind of a different name or DM process teaching people the skills needed. We have our own archive to make this possible and that's it. Hello, I'm BDL Garby. I think the reason Andreas asked me to join the panel is a combination of his desire to make sure that there was someone representing Debian itself, the main project that participated in this discussion and also a recognition of the involvement that I've had over the years in helping to foster this thought process that has led to some of the current ways that people interact inside Debian and outside of Debian in creating custom Debian distributions and the various downstream derivatives. If you go way back to the platforms that I wrote when I was running for DPL in 2001, 2002 and so forth, I was talking back then about how we could try to achieve this long-term vision of Debian as the universal operating system and even in that time it was very clear to me that the way this would happen was by presenting different subsets, different combinations of packages, different flavors of Debian in different contexts to different people with different needs and I'm very pleased over time that these ideas have been picked up and taken by other people and have helped lead to some of these different distributions that the others on the panel have been talking about. I myself have been involved in creating some subsets of Debian and helping other people to think about how to do that and to get started doing that in the corporate context. I mentioned in my talk yesterday morning the fact that Debian is the distribution that's used inside some products that come from HP and in effect the subsets of Debian that are part of those products are in some sense custom Debian distributions. So this is a topic area I've been thinking about for a long time. I am not currently personally today involved in any creation or maintenance of derivatives so with that I'll stop and we'll see what makes sense from a discussion standpoint going forward. I would like to speak also for the Debian MET project. The interesting thing in the Debian MET project is that there was nothing for no free software for Medicare in the form that Medics really could use it and so we were able to start it from sketch because there was nothing and we wouldn't be able to do so much work if there wouldn't be Debian and so we just learned that Debian is a thing which is amazingly adaptable to any tasks you want to do and we also learned that it is a so-called doocracy which means those who do the job decide which job gets done and this is a very good principle because if you can find somebody who is doing a certain job and he cares for this then Debian is a really nice playground to realize your vision about things and so we had the option to move this process very slowly forward. I wouldn't say that this Debian MET stuff is already a product. No, it isn't but it's amazingly growing and it has good chances to become a product. It's in comparison to the Debian Edu people they have much more developers and much more users because any school could potentially use it so the effect of Debian Edu is much more visible and it's even more increased because of the Linux people who try to merge and so you see the features which are there if you work together. This is one very important point, work together and try to realize your dreams to support your specific users optimally because I think this whole idea of custom-driven distribution is you have a special target user and this user needs some specific care for several reasons I will elaborate about this in my next talk so I want to go to much in this detail. By the way, I'm missing Thiago. He should also be here on the round table. I just noticed now for the Debian Brazilian guys, is he anywhere? I hope he's prepared, I'm sorry. Because I think it's very important in Brazil that people are doing an amazing job in bringing Debian to certain, Thiago can elaborate much more and it's very important that we try to find ways how people in countries could propagate the idea of Debian in this country. I know from, there is a country in Asia, Christian was there that is also applying Debian and we would like to know how can we as Debian make it even easier to roll out Debian in large scales? Thiago, are you able to give some words about Debian Brazil? The mic is coming. We have a CDD project in Brazil called BR Desktop. We are trying to put the CDD in Laney, I don't know if it's possible but there's nothing really special. We are working in Life Helper in order to get Life Helper working well with Debian Stahler and get the packages in Laney. I would like to hear from you, your experience. We have Debian E2 that for a long time has been a great CDD and in Brazil we started as a CDD just for a few months because I prefer to stay here just to hear you guys that have more experience in this subject. But if you have any questions, please be free to ask me. If anybody wants to comment on this? I would like to add what I forgot in the Debian E2 introduction because Debian E2 is Debian except for 30 packages we modified. We needed to modify a bit to make them suit our needs. The rest is pure Debian and the modifications are mostly configuration issues that we can pre-configure the installation so that teachers that it's suited for a school, the network is set up, prepared and it's easy to set up and that's our changes. We are mostly for over 99% Debian so what needed to be that we can use Debian even better is that for some packages modification is possible without changing confiles because they get overwritten upgrades but that's very few packages left so Debian is really well suited for customization already. What I also would like to say is if you try to make something else than Debian you should keep in mind or you should try to answer the question. What are the main three points that I'm missing in Debian? I'm missing in Debian this and this and this feature and that's why I have to do a derivative from Debian. If there are not three main points you should think about would it make sense to derive from Debian because it is always work to make some difference to an existing system so if you could elaborate on this one I think for the Librex people it might have been our government want a system which is ours This is one main point. Are there other points? At the beginning it was mostly the installer because we have end users who do not know how to use computers and they want to install Linux on their computers and it was really really hard for them to install Debian so what we did was to use a different graphical installer Anaconda probably you'll know about it but now with Debian graphical installer we are trying to develop the new release based on the Debian graphical installer but it also lacks of some features like a good partitioning tool and stuff like that and that is probably the main reason we do not use Debian itself So you found reasons and I remember this installer problem was also for Debian Edu and I think the basic work of the Debian installer was done by the school Linux Debian Edu people when it switched from Debian CD to Debian installer Is this correct or not? Not sure. I remember that Joey has some or whatever he was Yeah, I'm afraid I'm more or less the only representative of the DI team around so let's answer or try to answer The answer is yes The very initial work of DI was done mostly by Joey Yes, but the school Linux people and particularly Peter Reynolds and worked a lot, TOLEF also So yes, this is something where we can reward to school Linux and Debian Edu So this is a good example of a give back by a derivative distribution So the idea was I'm missing something in Debian and so I'm just doing inside Debian what is missing? So Debian had the profit and the project as well So this is the idea and I think this is the same those Linux people they learned that they need something and just did it on their own because Debian was not quick enough I think this is the main problem that Debian is a large thing and in German I've read the word Ehrenwoldig I don't know the translation what it means but you know what was... Honourable Honourable So it means it is a huge beast and it is hard to move it but it's not actually true I think You can move it and if you will not move it in the current release you can do it in the next release and we have seen that and this makes the success of Linux that they try to adapt much closely So I have two questions specifically to the drivers I guess I give them both but they both require discussion afterwards The first one is I didn't know that Debian edu is Debian with 30 packages modified and when I heard that I immediately had to think why are those modifications not possible to do inside Debian so that I could actually I could actually make the change in some sort of higher level policy file I understand that this is not currently possible that there are some conflicting changes and that you possibly can't answer them with Debconf but it seems to me that this is some sort of flavor question it's a group of decisions that have to be made together and then suddenly the system becomes a Debian edu system but I believe Steve is very keen on saying something to this so would you like to So I'm grinning and laughing at Holger when this question comes up because he and I have had many conversations about some of the changes that Debian edu has to make currently because the packages are not in a position where the changes can be made in a policy compliant way within Debian So Debian being the purest the technical excellence above all interferes with certain practical goals of being able to ship a system that does what they want and over time we are moving towards addressing, sorry, very vague so concretely we have things like Open LDAP is one particular package where there is no way for plugging in and the default configurations that we are shipping for the Open LDAP package in Debian do not accommodate everything that Debian edu wants to to do out of the box as an LDAP server as a directory server and we've been trying gradually over years to converge on something that meets their needs in a policy compliant way within Debian and it's a long process because they are hard technical problems to solve this for everyone and you didn't exactly want to wait three years for us to get that done before you were able to start deploying and therefore you did the sensible thing to modify the system on your own What Steve just said is one reason what's the problem is the other problem is the freeze that Debian edu has fewer developers so we can only start to deploy our modifications and test them properly when Debian is frozen but then we cannot get our changes in anymore so many of the changes we had were just minor bug fixes which the release team didn't want to have in stable when testing is frozen so that's why we have at the moment or two weeks ago we had two packages different from Debian we were pure Debian and by now we have, I have not counted probably ten packages different and I will try to talk about this more today at three and my Debian edu talk about the problems becoming main really So for me the most important thing there is that there's clearly a very strong commitment to get those changes into Debian and the way you articulated it and the way you, there's no doubt in anybody's mind that there's an intent to get that and I think it's valuable to respect that even when people have that intent sometimes it is necessary to have modified packages and there shouldn't be a social stigma attached to that as long as the intent is there to drive things in a positive direction I don't think it's a social stigma it's a technical problem we have problems but when upgrading we need to maintain the changes that's why I want to change it But I think in my conversations with customers and people that I work with and other companies and other projects it's always clear that what everyone wants to do is to focus on those small pieces of the development work that will help them achieve their specific goals and to help them in a commercial and business sense differentiate themselves from all the other people out there that they may be competing with and it's this notion in Debian that we do have a common place to come together to collaborate and work together on all of those pieces that don't differentiate us from each other and aren't unique to solving our specific problems and our specific needs that makes this whole community really powerful and as long as we all understand that and are trying to feed things back and are willing to tackle the problems of packages that don't provide the elements of configurability required to meet some specific set of needs then I think this is all very healthy and we end up with an overall ecosystem that's a good thing The challenge comes when for some reason or another some specific developer gets really unhappy about the direction that's being taken by a particular derivative and then we end up with conflicts or antagonism that at the end of the day doesn't really help anybody So the question for the technical reasons of why we're not being able to work closer together I think that's something that I would personally like to see addressed and maybe we can figure out what the question on a higher level not on a per-package level but on a higher policy level like what is keeping us from having a package that can actually serve multiple use cases I'm at the University of Limerick in Ireland in one of our software research groups they look at what's called software product lines and as part of talking to these guys I came across a study that was very very interesting groundbreaking that we all know on television sometimes they have commercials where they try to sell you games you can play on your cell phones so if you like this game press one if you like this game press two if you like this game and so on each one of them is five euros what I found very very interesting is that from this study or in this study they looked at these games and they found that most of these games are exactly the same game this is a feature file that you add to that or would you merge with it and then suddenly instead of having little green blobs run around the screen vertically you now have red squares that do the same thing horizontally new game new title done sold more money and this very much looks the software product lines people actually approached me because they are very very interested in Debian Debian being a system that has a rigorous quality assurance level for the metadata that we provide in our archive so that we can do our dependency resolution and also the cleanliness that our policy ensures on the file system level it seems to me that this is actually a perfect basis for something like software product lines is there a layer at which we can plug into the system where we can make higher level choices you know you can easily turn a Debian system into a completely different system just by specifying those few changes that have to be made I wonder how maybe a panel member has an idea of what to do there or what the technical requirements for that would be assuming that we don't have to we have our policy right now and we have Debian as it is right now but that doesn't mean that we have to have that same thing in five years from now so where can we go actually there is a thing Debian people are doing they are just wanting to put a different option to the installation CD and then it installs Debian edu and this is the final goal if I'm not completely misleading so they are constantly trying to get the diff between Debian and what was Debian edu very small and if I understand you right Florian in principle it's the same idea in Munich yes absolutely we've got some more changes at the moment but of course we want to minimize that we've had some dependencies in the past simply to open office because our serial system the Walmux you probably heard about that or Perps not it's free software for two months I think and it just does work only with the sun Java virtual environment so we had to use upstream open office self compiled so that's one point the other point I don't see a way out is new kernel for new hardware we have to build our own kernels and that's a very tough point we've got several departments and all of them are waiting if we say you got me great now and they say okay now I buy new hardware fine works great so this is one point where I don't have a solution in mind but in principle you're right absolutely so my conclusion would be that Debian is not flexible enough to solve this hardware problems and so on I think it is a task for Debian to try to get this settled somehow but I think there are some chances we have and Peter do you want to elaborate on this topic? Well it's actually Petter Reinholtz who's been commenting on IRC I don't know how socially appropriate it is to be watching IRC while sitting on a panel but he points out that it is actually possible to install an older version of Debian Edge using the normal technologies but that you know there's the package lag situation in the Lennie archive right now is what keeps the current version from being installable so it's interesting in lots of the conversations that I've had with the Debian Edge folks and with other people sometimes the issue really comes down to how we're going to structure the configurability of the package and whether there are in fact sort of large level configuration things that have to be decided at package build time that you know force decisions in one way or another sometimes those are solvable by being willing to have alternate versions of the package built in the archive that you know work in different ways this could be as simple as things like the Sudu vs. Sudu LDAP versions which was originally done because some people really wanted Sudu to work well with LDAP and some didn't want to bring in the huge pile of library dependencies that the LDAP inclusion forces on Sudu so you know there are things like that that we can do I think Petter in particular has done a lot of work over the years in how we should layer configuration file changes and make it possible for people in different levels of the distribution and derivation stream to be able to overlay different configuration information this is a topic that I'm confident could continue to be worked on over time to the benefit of everybody on some of these packages I don't know Holder if you have any particular thoughts about things that right now are challenging to configure that maybe we should work harder on we have our Buc 311 888 no that's the wrong number but whatever this is the Buc which blocks the upgrade for certain packages in DB and EDU those are our most pressing issues the Buc is about you should not modify configuration files and it's against DB and EDU config and there are 3 Bucs against DB and EDU config so it's easy to find the other thing what Florian said that it's hard to deploy newer hardware we all know this, this is a problem in Debian, Debian has the same and this year for the first time we tried to address it with edge and a half which I think was a good experiment took too long in the end but it was still there and we have started to establish processes for doing so and I'm quite positive that it will become will be better for linear and a half I think that will be done and I think that will be done faster so I hope this will be addressed better the other change you spoke about changes in Debian which we could do to make derivatives lives easier would be at the moment Debian is frozen but there are some packages where the freeze is not really so much in my point of view which is mostly the Debian installer it's differently frozen it's also frozen but it's different and I think it would be good for other installer packages to also have a different freeze guide lines like what I said Debian Edo mostly needs to modify three packages Debian Edo, Debian Edo config and Debian Edo install so that we can tailor Debian to our needs and if those three packages were excluded from the freeze it would make our life much easier and we could probably release within Debian and the same is true for Debian made probably for other custom distribution it's also true for installation thing which is also just takes Debian and is an installation tool so I think if I'm started to think how policy needs could be changed so that we can get this possible that's a very interesting idea so the thought then is that because those packages only really affect that particular derivative and are not things that other things within the archive depend on maybe they could freeze later in the process or much closer to the release time in the same way that Sledge was talking yesterday I guess about the situation with Debian CD almost always has to be one of the last packages into the archive because you know you can't really sort of do everything right with it until everything else is frozen and ready to go and I know in the past release notes have been some of the last things into the archive and I talked a minute taking a Debian history package update once very late it is after all just text and so this is very interesting idea and I think it's something that we could have a useful conversation with the release team about there any release team folks in the room that would like to comment about that one way or the other they're all hiding off doing a release thing so I also had a question to Mark because in my opinion I was asking for three reasons to derive I think in my opinion it is a very good job in branding Debian I know it from my institute well the story is a little bit longer we should use free software but there is no use except my private box there and I told them I don't care about the Linux distribution you choose this distribution which your next friend is using and I'm the only friend in the institute and so my advice was I would be able to help you with Debian perfectly and they decided no we are using Suzy because there is a large company behind Suzy so branding Debian is a very important thing in my opinion then you told us you are doing experiments and whatever it is interesting it was a new point for me and what's missing in Debian for you and what's the only third point you are missing in Debian what makes a reason to build your bundle ok so you made a couple of interesting points there and then I'll come to the things that really inspired us to create a derivative so I think your point of choosing the thing that your friends use is very sensible because you are going to get much better support from your friends than from somewhere else so I would agree with your way of thinking there for us the number one thing was the focus on the desktop and the commitment to a release cycle and the ability to provide commercial support so I'm contradicting myself there what I mean was branding branding and commercial report is I think to really deliver the benefits to a specific audience to a certain audience if we want to really deliver the benefits of free software to them then we need to figure out how to bring those things together and I felt it would take a dedicated effort just to do that and when you look at what we've done we've really tried to do that we've tried to figure out how you bring the benefits of free software which are the ability to modify it you know what those are this extraordinary long tail of innovation that you have access to and all of the communities behind that innovation to bring those benefits to the sorts of audiences that I really care about I thought it would take a dedicated effort that could make commitments about release cycles and could provide commercial guarantees and support and it's still unproven that we can actually achieve everything that we want to achieve but I think we've made good steps in that direction so our goals it was not so much a technical constraint it's not that I felt that we needed to radically change Debian which is why I'm quite comfortable with the idea that we constantly merge because we don't want to go off in a different direction it was more the sort of social and commercial structure around it that I thought would be necessary to bring free software free of charge to a consumer audience the question because you told me desktop and also Florian told about this Java issue I also face a problem that Java is not so good support in Debian as it should be because do you think this is a problem we have many applications that just not work because it's so complicated with this Java stuff I think that is changing and it's changing for a couple of reasons one has been that Sun's work on the license is bearing fruit now we really are at the point I think where you can expect to see a certifiable Java installation that's that previous free software and the second thing is you know we have engaged with Sun when I say we I mean Canonical has actually engaged with Sun and Sun is receptive to the idea that it's actually worth really packaging Java in a way that really makes sense to you know from a Debian perspective so we really want to try and create the sense of having something which is a true Java experience in every sense but it's also a true Debian experience that when you want to get this piece it pulls in that piece the Java pieces that you need the only the pieces that you need and puts them in the right sorts of places and my sense is that Sun is now actually really quite receptive to that and wants that and so Matias Matias has been working closely with the Sun guys and I believe collaborating with Debian Java guys as well and the intent is that we will actually have you know a TCK certifiable stack in Debian in Ubuntu certainly I imagine it would make it for Laney it should certainly make the next Ubuntu release and my always if I had some some Java application it's it was somehow a failure it doesn't work and so it's got somehow frustrating it would be really great if you could get some closer relation to Sun to work on Debian so this is a man okay this is great do you want to elaborate here on this I mean Doko knows the technical details but this is one of the things that that I work on at Sun is trying to get our software package for Linux distributions and the ones that we've identified are the top four Debian Ubuntu, OpenSUSA and Fedora we've been working very closely Doko's been a great help with this and also we have a person who we've recently hired Dalibor Topic from the class path project some of you may know him who's been working on this as well but we will be in we just got into main for Intrepid and looks like we're almost in Laney I think you just have to upload something we were Doko was talking with Mark Heimers yesterday about this and so it looks like we're going to make it in but that's something that's very important for us and then on top of that we'll start with OpenJDK and then we would like to get things like NetBeans it's one of the things I like to talk with Daniel Bowman because I think he's the person in Debian who's been working on NetBeans but there's other Java based things like our NetBeans GlassFish, our app server and things like that so if there's anything that you want to know about that Martin George, I'm around so you can grab me and we definitely see the value of getting Java into Debian and want to get it in there as quickly and as robustly as we can that would be really great so I've seen we have 10 minutes left are there any questions from the audience or to make not only a table discussion but perhaps there are some further questions if not I would like to come back to our DPL wants to join Hi, just out of curiosity how many people do you have working on HGO projects? I'm answering for myself by the way I'm elaborating on this and this afternoon for other CDTs and I think we have 3 to 5 very active people and about 15 people are providing some very good work so it's a very small project for my case Debian Edo, it's more like 5 to 10 people working on it, we have more translators but the core developer team is really small at the moment in the Linux project it's people working in the development team it's in ITIL terms spoken one change manager who isn't developing it's one virtualization guy who is sitting in the back of this room just right behind you and two developers including me but we're getting additional 8 until January next year so hopefully this will get better we are currently 5 main developers we also have some more people working on documents and also some of the teachers of the schools in Extremadura do some nice work and report it to us so we can be like I don't know 20 people and maybe 10, 12 teachers depending on how you measure it from about 100 and something developers who work on Ubuntu active on Ubuntu up to 20,000 people who've contributed translations to varying degrees of dedication or quality and I think we're very particularly strong in things like documentation the Wiki, the forums and communities around that so thank you for some sense of the shape and considering VCS Package is rather young and a lot of it is discussion I guess the number of developers in VCS Package is the number of people that take part in our discussions and then I would have to say maybe around a dozen but we do span 4 distributions at the moment so unfortunately Debian is definitely over presented so if you know anyone that is from other distributions that are interested in this make them join Steve would make them all Debian people if they were from other distributions I've gotten off that perhaps you have the wrong audience to ask for other distributions in this and that's from an Ubuntu guy Mark says any other questions? okay as I told in the beginning my main point was to try to specify the problems you have with Debian but why do you have something else and why are you doing whatever I also told the people from Debian Brazil just try to find out what could be changed in Debian to make it very very small or just be Debian like Debian people want to do and because it just makes more sense to save the work for different people this is my definite experience after 6 years in Debian we wouldn't be able to do anything without Debian this project would have died after 1 or 2 years if it would not be on the back of Debian which is a great community and really helpful to keep a good idea alive I think if we have some comments in the end then we could do so and then we have 4 minutes left I think I just want to make a quick comment returning to the technical stuff from earlier maybe the derivatives and specifically Holger because we talked about Debian edu earlier but this applies to every other derivative could try to think about what are the classes of modifications that have to be made to the packages maybe we can come up with a way that we can formalize those changes that those changes can be done easily in the package and I am specifically thinking of another distribution in the Linux market that has flags which you can set during compilation that gen2 use flags those basically are standardized across all the packages and if you want postgres support then simply you pass that flag on to the compilation of all the packages maybe we can find a way in which we can specify an interface at this level of packaging which the derivatives can use to simply create derive distributions distributions for other goals without actually having to go and spend 3 years getting depth con questions into this package and spend 2 years trying to figure out which package modifiable lift it up one layer and try to find the solution up there maybe it's possible from my side two primary things that I'd really like to see one is more adoption of packaging frameworks or systems that encourage split out packages and it doesn't really matter which system anything which explicitly says here's the set of changes that we've made in the process of packaging this is enormously beneficial in terms of relationships with upstream it makes it so much easier for them to see what you're doing and why you're trying to do it it's enormously beneficial in terms of derivatives because it creates a natural place where you can say well here's the additional work that we think is valuable and we see for example in all of the automated tools that we have where we're working with the package which automatically splits that out where the package structure is a split out structure the feedback that we can give to DDs or to the corresponding DDs is just that much clearer because it's logically laid out so that would be the technical change that I'd like to see is more adoption of packaging with split out patches and the social thing would be a positive engagement a sense that it's a really great thing for Debian that we bring more and more people in even if they are pursuing things that are potentially specialist agendas regarding the social and communication issues I think that developer gatherings are really really important to improve Debian thanks for extra Maduro for doing this you're welcome and thank you for example the Debian installer team was there in 2006 and they really really improved the graphical installer that's the reason why the new version of Debian will have a good graphical installer and that is also good for us I think that developer gatherings are really really important to get rid of the social and communication issues I also think that just meeting and talking is in real life or in cyberspace somewhere very important for us for us Derivates to stay in contact with the community get our changes back or get new things into Debian I think first step forward was the box-pushing party we did two years ago in Munich I'd love to see that again and we'll be trying to get new developers or at least maintainers going from Munich to give something back we're just starting so if you want to sign my key just get back to me that's it, thank you Martin spoke about a standardized way to configure packages and as there is nothing at the moment Debian Edu puts a lot of configuration into LDAP which is useful for the services we deploy but it's not useful for desktop applications it's nothing and that's from OpenWFT you start developing a UCI universal config interface which they want to plug into every package some I haven't really looked at it but the idea to provide a major level of configuration which many packages can use is useful but it hasn't really been developed except this UCI thing which I would suggest people to look at if they are interested in that it's an outline support for a enhancement to D package that would give us variant support within the actual package building itself and you would be able to leverage the full power of DEB build options and other standards like that to actually apply those at the build stage so you would be able to either have a slightly different D build package or to apply a different configuration to generate a spin-off package for yourself but at the moment it's just a bug report against D package but it is something that MW would like to actually implement across the board to provide these kinds of extra layers of configurability on the existing package not necessarily to have to rebuild it but to actually allow it to be built in a customized manner that sounds very interesting and something I think that it would be great for us to have some other conversation about getting the hook though we are out of time I would just like to close by saying that I'm incredibly pleased and just sort of amazed that in the time since we first began talking about flavors of Debian and how the project should interact with derivatives and so forth many years ago that we've now gotten to the point where so many people are so successfully creating custom Debian distributions and derivatives and that they're able to do this with so few people and I think it's a real testament to all of the work that's been done over the years by the huge number of people who've contributed to the Debian project and I think we should all remember that and think about how proud we can be that other people are able to take what we've done and go extend it and customize it to meet specific needs as easily as they can and remember that when they come and ask us to make something a little bit different maybe make a change and enable them to do something else that this is a good thing and we got to figure out how to make Debian an even more universal and useful base for people to do things with in the future. Andreas? Thanks for joining the table we have to finish here the time is left over for a couple of minutes so the next speaker is waiting. Thank you very much.