 Hello, everybody. Good morning. Well, my task is obviously to wake people up in the morning. I hope you had a good breakfast and slept well this night. And I would like to talk a little bit about the Deviantmade project, but it's not really directed to medicine, biology or what we are doing. It's rather I would like to give an example of how the Deviantmade project worked and what we possibly did right as I'd like to show. How many people are considering themselves in the Deviantmade project here in the room? One, two, three. Who has never heard about Deviantmade before? Who has never raised his arm now? Oh, okay. Thank you. I will tell you a little bit about past, present and future and I think of the Deviantmade project for sure and I'd like to outline what I think could be a model for other projects to get a similar success. Maybe you have read in my abstract that the Deviantmade project even is quite unimportant for Deviantmade has gathered one developer per year who confirmed that there are only developers because the Deviantmade project exists which makes more than 1% of the Deviantmade developers at all. But how did the project start? Basically it was an idea driven by some wine, some good Bordeaux wine at Dept. 1 and the question of the day was how to translate the installation instruction for some medical program and I think translating installation instruction is quite stupid. We have a packaging system and so I decided to have my first talk about packaging and I'm sorry for this interruption here and this was hold on the next day and after I had this talk I thought well let's make a real project which I started in 2002 and in 2015 so more or less 15 years after the idea was born I told to John Mattercore possibly everybody will know him that after 15 years I'm working for something people are starting using it and it was a little bit that I was not really happy about that it's such a long time and his answer was only 15 years you young guys should be more patient I should hurry up since after 15 years I might be dead but you have so much time I think this is kind of sanity and helps if you think my project is not going forward as fast as it should and it made me think that we somehow have done something right and are even right in time so what was the motivation to create a Debian Made project? Well in Debian it is possible we have doocracy I don't know if everybody is familiar with the term doocracy it is if you do something you decide what gets done and nobody will stop you and this is kind of good principle because it leaves a lot of freedom for the individual and you have a good idea and if you are very motivated to do it and you really want to get it done then you can get it done What is my view on Debian? Well you see a user has a computer and wants to install a lot of packages a lot of software and the software is mostly done or was mostly done by one developer you have a package developer relation my first package was WordNet I will tell about Friday on this one but the thing is we have in Debian lots of these packages I don't know the current number it is 50,000 or so so if you are a new user and a specific user you are really confused about all these packages how can I really find the package I really need and the answer to this in my opinion is so called blends we have a kind of a scholar linux Debian Edu blend, we have a Debian Mail blend we have a Debian Gist blend we have more than this but the thing is the blend gives the user a focus on the package he really wants to use so he doesn't need to seek all the pool for lots of software he can just use what the blend is presenting him and the blend is nothing else than Debian but this gives you a different view on the package pool a more simple view for the user who has a certain interest in some topic for instance, if you consider the usual biologist who wants to do some DNA sequencing then he installs the so called biological task of the Debian Mail project it's a middle package called mid bio and this has dependencies and he just installs a single package and gets all these packages and even more and in the Debian Mail team we have also maintainers you see my face quite frequently it's also not really really what's presenting it because in the Debian Mail team we have more than one maintainer per package in the good case or at least we are gaining for more people per package than only one to make it more reliable so this is basically the view we can provide and for every other topics we have similar meta packages and so people with other work interests can easily install these packages inside the Debian Mail project we have several so called tasks you here see the biology task with a lot of packages more than 500 you can do development of biological applications this is not so high bar but we have more applications and development support we have a little bit to support epidemiologists we have very very few to support hospitals HIS hospital information systems I will tell you a little bit more about this we have quite good coverage for medical imaging and development of medical imaging applications we have two packages where doctors can maintain their practice a little bit more viewers and so we have some packages to work in psychology it's basically maintained by the neurodebian team and we have some tools but the very interesting thing in this graph is if you look at the first five years well it's not really gaining traction it was nothing though maybe the thing is that I'm kind of an endurance sportman and have some long breasts to work with things which are not really developing and really had hope that it will work and the point where it rises that steeply is that we are able to form a good team which comes from different institutions have different interests and are packaging what they need and this is the point of Blenz in general we want to collect all people with a similar need in their work and they will concentrate on things to do something in common to make for all of them a good system and this graph is also a little bit reflecting that once we have met it was in 2011 we are doing sprints the slope is even more steep so sprints and meetings are quite important I see on this one graph I have other graphs which are showing basically the same so in short you need to be some kind of endurance athlete to do a long-term goal you need to find friends in Debian and you should meet this graphic tells you all this and what's the definition of a team for me it's just waking up in the morning I just realized that somebody else has solved your problem from yesterday I actually experienced this not only once while sitting down on my computer Saturday evening getting tired writing email to the list opening my laptop in the next morning while laying in the bed that's my habit sorry my wife does not really like it but I'm doing it anyway the problem was solved and pushed in the version control system and this is something I really like and if you are at this point then your project is evolving it took about 10 years, 12 years to this point so I'm just telling you because if you know that this might happen you can gain for this by the way you can always interrupt me in my talk just a small group raise your hand get the microphone and ask what are our ways to get a stronger team I think a basic point is to be a Debian developer you need a certain knowledge the knowledge about packaging is actually not really something secret it's if you can code write a made file that's basically what we are doing we are doing some scripts so any programmer should be able to to build a gradient package but this knowledge is it's good documented but you know who reads documentation this knowledge should be provided more easily and my idea was we can do the following I'm running the so called mentoring of the month and show that Debian packaging is easy I have some constraints I request open discussion I do not accept private emails the idea is other persons have the same problem and they can read it on the public mailing list and solve it without asking again and I also think that I'm not the only one who knows good advice and other people might give better advice additional comments so open discussion is a requirement and we work together for about one month it might last longer it might be shorter and the mentee can ask any question there is no such thing like a stupid question and even if it was asked before or there is a FAQ or so we will not punish this it should be very relaxed for the mentee to enter the project so the mentee is kindly invited to learn packaging in a friendly environment it's important I do not remember real flame bars on our mailing list maybe we are too few we are about maybe 50 people on the mailing list but it's usually a friendly list and we also are trying hard to attract upstream since the outsourced of some software are in principle the best packages because they know the software they know where the packages where the files should be directed to they know some configuration stuff and so usually if some upstream approach as could you create a Debian package of my software I usually invite him to do this mentoring of the month for two reasons one is that everybody has this to-do list and it's hard to put something in between so I usually can't immediately solve the problem and create the package so this is one point and the other point is if I train upstream to do the work he will continue to work and the effort I put on teaching in the long term be rewarded by not being forced to do the work over and over again the conclusions from this mentoring of the month are as I said the time I spend into mentoring is a very worth the effort I do not really have a student each month so the workload is bearable it's maybe three to four maybe five students per year and it depends how involved the student is how much effort I need to spend some students are very quick and I do not have much work and sometimes the packages they are trying to build are a little bit harder and it takes longer it depends it's also good for me to teach the students because I have read the documentation about ten years ago in the best case they have read the really recent documentation and I can even learn something from them that's good for me so I would invite you all to teach some newcomers to learn the current stuff the major advantage is that training upstream is somehow pulling the knowledge about the code and our packaging and so this is very efficient as I tried to explain and actually 50% of the students have some strong connection to upstream either upstream themselves or they have a very good connection and they are known in the bug tracker and can provide patches and so on and another 50% is not all people who are joining the mentoring of one's stay in the project this is also a wrong expectation if you get about 50% of those people that's good and then we are doing sprints sprints are meetings and Davion is also fun you know Gunnar Wolff, he's also organizing he's quite a known person and I once wrote him that I didn't told him that we are doing this sprint and my punishment was that I need to stand for 3 minutes in the 5 degrees cold water but next time we will tell him before and he said, oh I understand I thought Davion made focus on packaging medical software with his note you mentioned and your choice of location however I understand better after meeting every year you need medical attention possibly including a psychiatric evaluation but anyway so he invites us to Mexico with better conditions yeah maybe he's right but usually I say in some aspect everybody here in Davion is a little bit crazy it's different than the usual person on the street, that's fine my wife considers me as quite normal in this scope of the Davion developers that's a compliment so about the team well I want to say more about the sprints we are meeting once a year and this happens mostly in wintertime on the beach because hotels are cheaper there you should make sure that there is a good bandwidth connection which was kind of not working all the time and we are about 10 to 15 people there about half of them is constantly coming Sasha is close to always coming and other people are coming only once and don't remain at the project but that's not important we are meeting and as I have shown on the graph before the meeting are really pushing the number of output we have so this is really important and Davion is supporting these meetings we four or five Davion developers who are usually attending we get the travel course refunded and thanks to Davion for this the funding is partly also done by local organizations we were once supported by the university of Ren when we did this meeting also you find some sponsors this works also somehow and as I said it's quite important yeah this graph really shows what team maintenance means it is not as positive the outcome as I would like to have it but I will explain it let's not switch on this bar shows the number of packages which are inversion control system only touched by a single person I would like to have this graph drop to zero say well every package is touched at least by two persons this would be the goal we are far away from this goal if you look at the comparable graph of the Perl team the Perl team goes like this this is the goal we are not there and even this bar says well two people are looking at it but well if you know the interns it's in frequent cases say half here is somebody started the package and I took over so I'm the second one which yeah this person vanished and I just keep on working so it's we can do better here but anyway we have packages which are maintained by three to four people that's the start we try to get better I create this graph in so called team statistics project you can have your graphs for several teams and just ask me if your team is not amongst it I can create it it's quite interesting because we have some teams where you have only one bar and then it drops very very strongly so just check it for your team now I come to a point I'm very proud about because I did some questionnaire we have a group of developers other deviant developers or deviant maintainers deviant maintainer concept is known who doesn't know what is a deviant maintainer deviant developer like me deviant developer like me and such and so we are official members of deviant we have voting rights and we have upload rights the step before is not the step before if you are not a deviant developer you need to ask me I have prepared a package please upload it now I will do it for you that is called sponsoring and assume I have sponsored you for two years and I know your package is okay and I'm bored by always looking at it you can become a deviant maintainer and you get white listed for this specific package and get upload right for your package this is a deviant maintainer and so we have I don't know exact relation maybe 28 dds and 2 dms also the deviant maintainer status is in most cases the step before you get an official deviant developer because once you are a deviant maintainer you want to have right and become voting rights and so so this is the team and I was asking myself what attracted the people to come to deviant I have also from my other talk on Friday I just wrote to the mailing list please write down your motivation to come to deviant and I asked the explicit question who is a deviant developer just because deviant meet exists so if deviant meet wouldn't exist would it be deviant developer or not interestingly 16 of the dds confirmed yes if the deviant meet project would not exist I would not be a day-to-day because deviant would not be so interesting for me and while these 40 dds were there before deviant meet existed if deviant meet existed before I was in deviant I would also have joined deviant because of this what does it mean actually if I'm honest deviant meet is not very important for deviant itself if there is biology stuff and medicine stuff inside or not it's not important redhead hasn't it in this amount people are using redhead anyway so the deviant meet project itself is not important but this quite unimportant project has gathered 16 developers this is more than 1% of the deviant developers so the lesson we can learn here is if I have an attractive project which fits my topic I can make deviant way more strong than it is currently so the main idea of my talk is please create a blend around your topic and gather developer for deviant and it's not only in the project because 10 of these 16 people extended their activity to other fields in deviant so they came via deviant meet and now doing other very good things for deviant I think this is also a very important thing I do not want them to stay here it's also doing other stuff so it's important for deviant that we try to get new developers and not all of them are active in deviant meet anymore only 13 of them drifted away to other fields and it's also perfect so again make deviant attractive for a certain user group and you draw from their new developer power who is using deviant meet but the short answer is we don't know we can't know, we can download it from the internet that's a typical thing for deviant we just don't know we have some popularity contest popularity contest is if you install deviant you get a question, do you want to submit data about your packages you are using to some server on the internet yes, then the data is published and please do so we are not spying your computer it is encrypted, it is anonymized and the actual data is not visible for the public but only the aggregated data so if you say yes for the popcorn we get your data and we know what packages are interesting for you and we put more effort on these packages that's for your profits so most probably we have more users than popcorn numbers are showing since the default value for popcorn is no because we care about your privacy but please, this is probably the only default value you should switch on the installer please switch it from no to yes and we also don't know how many users are using our work via derivatives as far as I know in biology Ubuntu is quite popular so probably ten times more users are using our packages via Ubuntu this is fine, this is okay I will not blame anybody about using Ubuntu my opinion is that there is some drawback because in deviant we are faster we are a team behind these packages and we are doing backpots and you would get way more fluent and way more current packages if you are using the biology packages via deviant so use whatever system you want but my suggestion is in this specific topic deviant is better for you I had some example a former colleague of mine needed some package I did the packaging with two reverse dependencies everything was done and I asked him are you happy now it's not available for Ubuntu I said well you need to rebuild this only I can't do it, so the question is if Ubuntu is easier to use, I can't decide but if it fits your topic that might be interesting and we have other derivatives we have for instance biolinux biolinux was an Ubuntu derivative we are doing some sorry doing some interesting stuff more visible on the desktop and some adaptation of the desktop and maybe some additional packages but the good thing was the biolinux developers were joining our sprints were using our version control system and so they put their development effort together with us and this was really good and so they developed in Debian and then built their system on Ubuntu which is fine sometimes we know about the usage if there are some publications there was some publication with the quotes welcome trust Sanger Institute maybe the largest institution in Europe which is doing bioinformatics and they have a really really large system running at this time they called it the biggest Linux web partition and they were using Debian so even two or three Debian developers worked there not really using Debian Mids packages as far as they were not Sasha was working there and he was but before Sasha there were also some people I forgot the name they were joining Debian in Edinburgh but they did not really join the Debian Mids team they did other stuff maybe we were not as good as this time I don't know so if I'm talking about Debian Mids why I'm talking so much about bioinformatics the reason is simple we have most of the free software in the field of bioinformatics when I started with Debian Mids we had in Debian 5-10 packages mostly offhand this was my start to de-offhand them and update them and take them over so bioinformatics is very good covered in Debian Mids and that's in Debian because we are in Debian we have a very good we have also a good coverage of medical imaging something is missing which we should package which is quite complex with Java stuff non-free licenses and need to be freed but this is also something we are actively doing we are approaching upstream and please use a free license we have also a wiki page we have a list, we have succeeded with this and we have work in progress this is one of the most boring jobs you can do in Debian but it's somehow needed and the good thing is if you are successful and have a list of success it is more inspiring for other people to do it as well we have something for medical practice it's called GNU-MID and free med forms GNU-MID is basically developed in German and free med forms in France it has some yeah it implements somehow also country based healthcare system so GNU-MID is internationalized, should be work in other countries as well but it's definitely nothing the general practitioners should just install the computer and start running this practice the general practitioners should care for the patient and not for the computer should hire an IT person so it's not that bad that we have only two because if Debian provides a solid base with Dicom viewers which are used in the practice and several things are packaged not actually the management system but something which is used in the practice it is also some help and for hospitals we have, well I would say close to nothing we have one thing with this FIS GTM it is kind of a database which is used by the system Vista Vista is a very old hospital management system way older than before Windows Vista and so this is the only very powerful hospital management system I know and we have at least the database it's running upon the system itself is so complex that people told us it can't package it I'm not really convinced but I also can't do the work and improve them wrong it's it's hard to do and the good thing is we got this FIS GTM in the first mentoring of months project because it's also very complex and has a very skilled mentee and we worked very good together and finally we have at least this and what we can provide to hospitals is a very stable and reliable operating system not specific with their adaptations they need but we are the universal system right well when I first mentioned the name Debian Meet people immediately assume I have the prejudices that we are developing code to maintain patient data I'm sorry that's not the case we just provide what exists as free software and do it as best as possible so if some free software exists to manage a medical practice then we are packaging this and help you and as I said the doctors should always consult an IT professional to run their practice there are some problems with certifications in certain healthcare systems but there are also solutions for instance I mentioned this in the first talk here in the on the public day because taxes are also I don't know was it in Germany or not taxes should be provided by some proprietary software to not expose the data to somebody who could hack into it this was what people said and say well that's actually stupid because in the patient data should also not manipulate it and the solution is the system creates a hash zoom over the data set and moves them to a different server a notification server and there it is stored with another hash zoom cryptographic ensured and so you can always prove that the data were manipulated or not so the certification authorities should think about not hiding the data to not becoming manipulated but making sure that it's be provable that the data are not manipulated and that can be perfectly done with three software well for the biologists you have some practical issues that biologists are updating their system to Debian stable about every second year and in Debian made we try to release software via back ports to have the current data the problem is one scientist needs a stable version for a long time and the other one needs way up to date or last published version and so we try to get some solution with containers, virtual machines also to install them as quick as possible and to make the access to this as fast as possible well the main point of my talk is in the last century we had a one to one relation, maintainer per package and then started team maintenance team maintenance means you have M maintainers more than one and one package but in blends we are creating maintainers for packages also an M2P relation it's not one to one it's any number of maintainers with any package so it's rather a team to topic relation and this actually makes blends in my opinion way stronger than any other approach to the package pool and also this makes blends the missing link between upstreams and users of some specific topic this is the main point and my opinion is the point why we are able to gather so many people so now it's up to you we had the talk in 2013 Ashish Lorayar said it's in the video about minute 33 we should try hard to run around asking users and developers is there any topic you care about create a blend today I asked him two years later if he was running around to create no he didn't but maybe you will so if you want to like this team topic relation create a blend today and I will help you here at Debcon for remotely and so maybe even you are able to attract one DD per year just by following the Debian made example yeah this is what I wanted to tell you the talk is available here you can also Google for Andreas Tiller talks and you will get a link are there any questions is anybody running some or working in some team with a certain topic which might be interesting do you need more coffee yeah if we have not packet already in Debian we could start to work in Debian made since the beginning without any knowledge about Debian packaging and we could learn here yeah that's the goal so assume you are you need if people ask me how can I do Debian made I ask them what package or what program are you using which is not yet packaged if there is such a thing then let's start packaging this one because this is some software you are interested in and you are catched more by something you are interested in than by some random WNPP package right thanks so so I am not a DD or anything but I am planning on getting involved and I am wondering if you see any difference between sort of explicit teams in main like the Perl team or the JavaScript team or whatever and blends well no there is actually no difference it's just a topic what they are doing the topic of the Perl team is Perl libraries and the topic of Debian made team is it's just technically harder because we have to deal with all languages and Perl has a focused developer team but we adopted for instance the policy of the Debian Perl team we adopted it and used it and tweaked it a bit so it's very similar but we found a different name to distinguish from technical teams but there is principle no difference does anyone have another question if not thank you for your attendance see you next time tell me if you created a blend thank you Andreas thank you