 Well, good evening everybody. We are about to hear a talk about the VN Science by Andreas Stilde. As always, if you want to make a question, just raise your hand, wait for the microphone, and if you can tell your name and speak to the camera. That's all. Thank you. Good evening. So I'm the speaker who has scheduled two events in a row, or you might have seen the video, I'm keen on seeing it again, but now I admit I'm really refreshed to hold this talk. Before I start, I want to say something about my t-shirt because it's no offense to you. I don't think you are morons, but I always try to use this t-shirt when I'm holding talks to make some reference to the story of this t-shirt. I was joining Davian in 1998, and after two years, our well-known and famous G-Lib C maintainer wrote an email, goodbye boys, I will be not alive in about two weeks anymore. Nobody was aware of this, and he was ill, he was 21 years old and had a disease which just removed his muscles and so on. At this point, I was feeling that I'm a member of a family and not only just some random geeks which are building operating systems, so I'd like to choose this t-shirt to remind you on this thing. Well, okay, now to the talk. I got one question in advance. Is Davian the distribution of choice for scientists? Well, the answer was no. I want to describe the process of turning Davian into the distribution of choice for scientists. And I would like to try a new term on this because formally we had a different term for this stuff I'm talking about and I would like you to think about whether the Davian integrated solution is something that would fit what I'm describing here. I would like to pronounce it dish because we provided dish with a certain flour to you and the flour in this case is science. So, we have some scientific, specific dishes in Davian and one of them is if you are working in the field of biology, you can use the dish Davian made and if you are working on geographic information systems, you can use the Davian Gis dish and there's even for chemist, chemist the Davian project which tries to work for chemist. Well, but for the others there is not so much. What are the reasons why these specific fields of science are better supported than others? I wouldn't really say that they are better supported regarding the number of packages we have. We have several packages of scientific software which are really interesting but I would like to propagate some kind of formalism about these packages because I think building distribution is not only adding packages to a package pool. We do very well in this. We have 22,000 packages in the package pool and it makes no sense to make 30,000, 40,000 without building some substructure. It just does not scale. And so the idea of dish needs some time to propagate or penetrate through the maintainers that there is something more to do than just packaging. Unfortunately, the advantages are not so widely known as I would like them to be because people are just making some fun of me. You are always talking about the same topic. But is there anything to talk about? And I said, well, not that much but it's growing and I hope that it's growing further and further. And most probably why is now much more of this dish is that it's a lot of work to run this and nobody really wants to do the grand work of this. So building some web pages, care for some wiki, maintain a mailing list and gather people who are involved in this topic. And so this is a lot of work which is not really fun. You have to write documentation. Who likes this? Hands up, who likes documentation? Come on, who likes documentation? Oh, well, about 5-10% likes it. Well, this is a problem. Programming is fun but running a dish and making it a closed system is really work. We do not have so many people who like to do this work. So the idea of running a deviant science dish is well, if there is nobody who cares for specific mathematics or physics for the moment, we try to find a common umbrella for them and try to penetrate the principle to the people. And if then comes somebody and well, now I want to have a really more specific field for high energy physics or whatever. I start with a deviant physics project or whatever the main will be. And so we could get more easily in this process. What did we had before? Well, three years ago at Debcon 5 in Oslo, no, sorry, in Helsinki. There was a first talk about deviant science and the result of this work was, this talk was, we had a mailing list with some specific user support and some wiki on web pages. But it was not really alive in the case that it was not really focused on this dish tools and it was more or less a select discussions of scientific issues in deviant. What program could I use to make fine plots and how can I do 3D plots or what program are you using to make some regressions also. And this was basically a user list, even speaking user lists where scientists were hanging around. Then we had some categorizations of scientific packages which was nice because I really like this DevTex effort which was somehow helpful to find those scientific packages you are interested in. But unfortunately this DevTex technique is not so well known as it should be and I have certain issues with the maintainability of this and well, DevTex is definitely a topic we should elaborate more about also in this dish effort and we have to work on this. I take vanilla and strawberry, if you like, they ask the color of the mobile phone. Then we had Debram, I just detected Debram when I was preparing my first talk about deviant science. I was not aware that there is such a tool, who knows Debram? That's also not that much, even less than the documentation writers. So Debram is also some effort to classify packages but I think that DevTex is much more powerful. My estimation is that Debram will die sooner or later. So what about the deviant science dish? I think a dish provides straightforward access to the package pool by providing some techniques which I have explained in my talk about the formerly so-called CDDs. So Tuesday and the interesting thing is there is no need anymore to search for the interesting packages. If you are a scientist and get some grip on the deviant DVD, you just put it in your computer, install it, then you get the task there, do you want the mail server, do you want the desktop environment, do you want the database server? Well, no, I just want to do scientific work, what now? And then you start firing up Synaptic or whatever package selecting tool and start seeking for software which might be fit for your purpose. And I don't think it's so much fun to read all these 22,000 package descriptions. It just doesn't work. You just want to know, well, I need a workstation for my purpose. And we try to provide this by telling the scientists, well, if you are a mathematician, just install one package, it is called Science Mathematics. And this has a dependency to everything which might be interesting for mathematicians. And this is quite easy to do. You can do it with one line or one click in your Synaptic or whatever. Moreover, we have some engagement for packaging more scientific software. Yes, it is right. We need more packages in this field. And we try to make a useful selection of the free software which is available. So it is also about adding packages, but not only. Moreover, we found some techniques to add some user menus or some pre-configuration if it is needed for specific software. So you can sell your scientists. Once this Science Mathematics or Science Physics package is installed, you get a separate user menu which is called Science. And then you find the interesting packages exactly for you. In the field of science, it is not as worse as in the specific sciences because the original Debian menu has also a science menu. So this is nothing really new here. But if it comes to chemistry also, you could subcategorize in this menu which is not possible in the general menu because we have the menu policy and we can't break it. And you could also do some adapted installer. The installer then does not present the task selection from the default Debian installer with this desktop system database, whatever. So what you can then select your specific science. And this does in principle the same as installing this meta package. You can click on the science you want to have and get the box ready and prepared for your scientific purpose. What specific meta packages we currently have? There is Science Astronomy, Science Biology, Chemistry, Monics, Engineering, Geography, Linguistics. It has not so many packages but we have this really nice package wordnet which is unfortunately not so widely known. Mathematics, Physics and Robotics. So you probably know that there are some, well, something which you call really a science and something which is not so called. We tried a pragmatic approach if you are working in this field and we have packages which are important then we add such a meta package. There are some other things missing but the missing things are well we have no packages in this category so we do not just invent a meta package to fit the scientific scheme. And we have some kind of common science utilities. Utilities for instance for statistics or type setting you will find all the tech stuff in it and BipTech and something what you need for your scientific work and just for viewing some viewers for scientific data R and GNU plots and something like this. So these meta packages are created by the task files which are the input files for the so called CDD def tool. We now here are facing the remaining of the old name Custom Data Distribution, that is not the name because you might not find the new name. And this stuff was basically done in the Extremadura meeting. I am just mentioning this because I really want to thank you the Chunter de Extremadura for inviting us and specifically Frederick Leo Bay and Dominique Behachemi who just did the real work, this grant work I mentioned in the beginning. So further we did some sensitization of the Deviant Science mailing list where you were just a group supporting your users but we want to try more, want to go more in the detail and become a real dish and provide some real solutions for the user. This is the main problem and there were some heavy discussions on this mailing list, do we really need this? This is a lot of formalism and our users are happy anyway but I think we have no real numbers about the happy users and the unhappy users and so I think it is better to try to do some more than just to expect that users are happy. So we have, for instance, we have popcorn numbers where we can obtain the numbers of people who install the specific software and who are actually using it. There is sometimes a big difference between the people who have installed the package and are really using it but these popcorn numbers do not really scale because scientific software is somehow specific and web browser is used by anybody or text processing system using many, many people but this scientific software has not so many users and so we do not know whether we have so many happy users or not. So I would like to present you some URLs you might like to look at. This is, in principle, the Debian Wiki of Debian Science and the second URL is even more important. If you would like to contribute to Debian Science, this link is for you and this Debian Science group has even a policy invented to make sure that packages which are maintained by this group are all in the same shape. In principle, we discussed in the meeting of Krieger Herrmann about group maintaining, how groups are working together and it is good if you have a clearly defined policy how the packages look like. And we tried in the Debian Meet a dish which is kind of an example which was taken over for the general scientific dish a policy which clearly defined what is the maintainer name for a package which is group maintained so this is a mailing list and then you have the uploaders and we have a common version control system which is in the case of Debian Meet and SVN system. Some people don't like SVN, they know that it is much more performant and they really like to take it and that's why we even decided in the Debian Science effort to use Git. This has some advantage because we can work together those people who have biological software and would like to work with the Debian Meet team but refuse to do this stupid old fashioned SVN can check in in the Debian Science SVN because it's also science and so we are working together and I try to establish this cooperation because it makes no sense to sit on your package and this is our package and this is your package we want to work together and then finally it's about users who just can use this package. And so the policy is technical, a little bit different but we have clearly defined things we want to do and also the policy contains some how to use this stuff but I'm also, I don't really feel old but I'm over the average of the normal Debian developer and I'm a little bit unflexible changing from SVN to Git it takes time and well I know something why should I take another thing but if there is some step by step introduction just do this and this and this this is easy to learn and so people could get used to it quite easily and it makes things much easier to get into the group and this is one piece of the documentation which has to be written and this is quite important so if you want to join the Debian Science Group have a look at this policy then we have this so-called task overview I just show you what you can see there here you can see the result of the definition of this meta package you see here astronomy and for instance I will show you the content which is just included in Debian so we have here some astronomical software and the green bars contain those packages which are inside Debian so it is quite a lot I formally did not know that Debian contains so much astronomical software and now we come to the yellow part this color below is yellow in contrast to the green above the projector is not so good in displaying colors the video is much better and the yellow stuff is just what we are working on so we tell our users well it is not just included in Debian but it will be in the near future and there is even an unofficial package and if you want you can install it we are not really responsible for it but we are working on it and then we have some projects on our to-do list which should be included next and this color is in contrast to the yellow above this is red okay and we have some WNPP bugs which means there is somebody working on it and we have even the person who is responsible for this and so we are tracking stuff we want to work on and if you are wondering how we are caring for a user V I can show you we are caring even for other languages so for instance if you want to see this in Japan I am really proud this works we have so many Japanese users who could browse here this one is not yet translated but we have a lot of translations and so you see on the bottom of this page there are some translated package description in several languages and what is also my specific duty is that if you think well this translation is not right then you see a link fixes the translated descriptions anybody speak Japanese and we could try to fix no I don't think so I will not touch this or if there is no translation available then you can translate this description I could use a different language and whatever now you can go to this translation screen and you can start translation translating this and my idea behind this is if you are an expert in a specific field you actually understand the description of the package I several types try to translate descriptions of biological software because it might be needed by some colleagues of mine but I am a physicist by profession I do not really understand it and you can't translate a description which you don't understand and so I try to involve our users in working for Debian so if these people translate the description of the package the next user is able to understand it in its mother's language which is quite important because we actually have a quite high language barrier and perhaps we can help a little bit in going over this barrier and this is what I mean with running a dish it is not only packaging it is also trying to attract users by showing them what we have and by making them understand what we have here also for chemistry if we are in Argentina we try the Spanish translation I never checked whether there is something there but I hope this looks quite English to me oh here is a Spanish translation so if you are interested go to these task pages click the links and translate this would be helpful so this was the task overview then as I said we did kind of we took Debian made kind of a model for this dish I will show you some other features which are not yet ported for all the other dish I will do so in the next couple of months because we had a very active developer who did a really great job in inventing these web tools and I am showing you for example this Debian made stuff which will be soon available for the others we have our group policy and we have the bugs so if you think that any scientific package has a bug you will find it on the list of bugs and so the people who are working on the dish can easily track the bugs which are gathered by all the packages which are entrusting for them so it is sometimes hard because these packages are maintained by a lot of different maintainers they do not necessarily have to be in this dish group and so it is a nice feature to have a good overview these are our interesting packages just let's see whether there is some bug or not and they are even marked with some color scheme which are not so important this is for the bug overview and then we have a quality assurance page and this shows how is the status of the package for instance we have some difference between the available package and the upstream package that means normally somebody has to do some work and build a new package so also in this view you get exactly the packages which are interesting for your target user group and have an easy access to the stuff you really have to do or you plan to do so this overview will also be soon available for the Debian science dish and on Tuesday I also presented some interesting stuff about the activity of a mailing list which is some measure for me which I think it should be a good measure to find out how this group is working and I did this measurement for the Debian science mailing list I do not really like that I am the first of the top 10 posters in the Debian science group because my plan was to be specific for the Debian maid group and you have to concentrate on a project and should not spread all your work on different groups and so I hope to be able to reduce my work but what you can see here is there was some certain work done in 2005 when the group was founded the interesting thing is Charles Plessy, number 2 joined the Debian project in 2006 and he is a very active person in the Debian maid group as well and I have told you Frederick Leo B who was working on the task pages is also very active on the group and so we are trying to form a strong team which stays in the group and works for the purpose that it is alive we have other mailing lists which have only a single person more or less posting and if this person comes for whatever reason idle getting a new job and have no time anymore the project is dying and this is very dangerous and so we have to try to find some means to attract people you see some new peaks Adam C. Powell was quite new to 2008 and he is supporting the idea of running a dish and well, I hope that we find some more and that this mailing list will evolve quite good the really large peak in 2008 is clearly only influenced by my effort to turning the mailing list into a dish and it is kind of an artifact and not that active like it might look like so as I told you these web tools which are provided by the dish framework which are influenced by the task files as a basic source of information are build it and the QA tools of the Debian Meat dish will be also used quite soon and so the final plan is to form an instance for upstream developers of scientific software what do I mean? I had made the experience that in the Debian Meat team upstream developers have no idea how to propagate their software to their users they put some random tar files on any accessible place on the internet, on the web page, on the FTP server and then this software is hanging around and does not find their users and so they are quite happy that we are caring about this and trying to build ready to install packages and this really gives the kicks for these upstream developers and so they are talking to us and talking to upstream developers is a really good thing we have seen in the other talk with SSH think that we probably do not really talk enough to the upstream developers and this is quite a good idea if we just to talk to them write some email you have done some nice software we want to integrate it are you happy about it? we would like you to change this and that could you prepare your software in a way that we can easier work with this and all this stuff and this is a very important task because I think that many maintainers just take software as it is and try to make a package somehow and the contact to upstream is not so good if you contact upstream with some deviant science team which is an instance and is somehow known then you have another position and the people want you to do something and they will follow your wishes much easier finally we want to work on the to-do list of the not yet included packages it is what I have shown you here this is our to-do list in the astronomy case the packages which are not yet included this stuff is kind of interesting because if somebody issues an ITP I intend to package this or that then my estimation is that half of them will not be solved by an upload but will be closed after one year because this guy has lost interest or whatever but if he is checked here then we will get a grip on it and will not forget it and so this to-do list is kind of important because there is so much stuff announced that some people want to do it and finally they will not doing it and so some other people could step in and finalize the work so what's on the to-do list well I think we should just go on with the nice work this stuff more or less started in the beginning of this year so it is not that old and it increased the cooperation the deviant may team a lot formally I was not aware that there is a real team but now there is so we want to increase the support of single sciences to create more specific dishes as I told you in the beginning well science is in a large field and finally it is not that useful as if you have specific science because no scientist needs all the available scientific software and so as long as there is nobody who really cares about the specific science we try to be the umbrella for them but if there is anybody just raise your hand and join this effort and there is much experience how to start this there is even a documentation how to start a dish which is quite straightforward and I think very good to read and we also try to advertise the support of sciences in deviant to scientists so these web pages we created are kind of a nice way to show off here we are, this is our stuff there are a lot of other scientific projects they just provide a list of package names or whatever but what is a package name if you don't know the package then the name will not enlighten you very much you need real description and we provide the description and we provide the description translated and this is a big difference to other projects and we obviously try to invite these derivatives who also build some scientific distributions but join us, we are open we do not want to try to reinvent the wheel we want to make just our users happy and not try to compete with anybody else who also do the good work surely not the only team which does good work there are many other teams and you find for instance there is this quantian live CD which is run by Adam Biddle who is also looking around on this mailing list and he is providing a DVD which you can put in your computer and which has a lot of the scientific software included but I think we could support him by doing his work by making it easier and this is also helpful so we want to help and would like to attack people to help our effort so this I made this talk a little bit shorter because I really hope that we could discuss a little bit about this stuff it's actually not a talk but a boff and I really would like to invite you to say your opinions and perhaps some people in the room who are interested in a specific science and just ask me you will find this talk as soon as I it is just uploaded but not linked from this page and I will do this in this evening and this will also be available via the conference management system so at this point I am finished and I am happy to answer your question now Hello, so my experience in my field which is Automated View Improving is that people who are working with these tools well most of the time need the most recent version which often comes even from CVS and not even officially distributed by upstream so they often take newest versions from CVS the newest features they often have demanded themselves to compile them and run them locally so the question is whether such an official distribution which tends to distribute official releases will be sufficient for scientists Yes, well there are two extreme situations I know other people who really want to stick to their old fashioned version because this one has provided the data which are needed so well to address your question it was a long time ago I think it was in 2004 in Malaga we were discussing about the technique to well it was not really an idea for realization but just an idea we are talking about released Debian not as a whole but just as such in the time called custom event distribution so make very specific releases to you have a running testing at every time put for instance scientific software together and have one DVD to hand out to scientists this is not implemented I even don't know if it is possible to implement but this is at least an idea which exists which documented so if you think this is useful to work on it we could think about this but it is probably a lot of technical work it is also interesting for instance if you are releasing such distribution for schools schools normally tend to change their software in the end of a year of school to prepare the next year and not correlate with the Debian release cycle so it makes perfect sense in several instances to say well we release a specific field of Debian and make not this release but if you would go to a mailing list now I think it would just end up in a flame wall it will die really soon you would have to present something working first and alternatively ok well if people have no fear to install Unstable we try to get the latest version into Unstable that's why we have these QA pages you see the moment here is the QA page you see there is upstream and we try to care to get the latest version and if it makes any sense to use SVN well we can try it just talk to us so maybe a more immediate solution with infrastructure that we already have would be to make extensive use of backboards.org but in my experience people may need not only the very latest to improve us but sometimes actually for versions that lag behind a bit versions that lag behind a bit because if you see to improve us as a compiler for a language the language that they accept is extremely incompatible from one release to the other so upgrading versions is real work on your whole library of proofs that you have already done so I'm considering whether for these we shouldn't actually package each version as a separate parallel installable package well this is not implemented but in principle I could think about that we could try to do this because also as always see this in this dish effort and it could be very sensible for other dishes as well and so if you think about this deviant integrated solutions we could implement a technique to work for instance with snapshots deviant.org and try to get a specific version from there I think this is quite doable I was really thinking of being able to install several versions in parallel with the same machine but considering doing that for Isabelle you could probably pin some version also I don't know I admit I was not very I have not thought very deeply about exactly this stuff but feel free to think about it whether you try to implement something anybody else? I'd like to advertise the next event from Munich which will be here in 50 minutes and thank you for your attention