 Hello everyone. Before we start, I want to ask you, this is above, so the idea is that you will have to speak. Please, when you have to speak, raise your hand, we will give you a microphone, look at the camera and then start speaking. Well, now I would like to present to Paul Weiss. He will present us bits from new maintainers and users. Paul? Okay. Firstly, I'm a little bit unprepared, so yeah. The full results will be released on Daven Development announced when they're done. Some of the issues that users and new contributors have brought up have been addressed since the surveys were sent out and responses gathered. So, what basically I'll be talking about, what I did, the user survey results, the new maintainer survey results, whether I'm going to do it in the future and then we'll open up for discussion. So, basically I did two freeform surveys, one of Debian users and another of Debian new contributors, the new maintainers, application managers, people doing sponsoring, people having their packages sponsored. The survey was specifically freeform, so people could answer however they felt, but I did mention a list of questions that people might like to respond with answers to. How many of you saw the surveys? Few. Okay. I specifically asked people if they wanted their comments published and or if they wanted them attributed to them. A lot of them said they prefer them to be anonymous, but said that was fine to publish them. The user survey was advertised on Debian user on forums.debian.net and also on IRC a few times. The new contributors survey was advertised on Debian mentors, new mate and IRC. Both of them were also advertised on Linux Weekly News. Had a couple of contributors advertising at them on language specific user lists. That was the user one. Okay. So basically what I did is I sent out surveys and sent out reminders occasionally. I replied to the males in batches as they came in, clarifying things, informing people of in-progress solutions to the issues that they found, counted some misconceptions. The analysis of the results was done during debcamp and debconf, which is why I'm a little bit unprepared. So basically what I did is reread all the emails, form sections based on the common answers, filled them with summaries of what people have said about the sections and some statistics as well. So the reason I did the survey was to connect personally with users, connect developers to the wider user community, to gather information about how Debian is used and the variety of locations and things that is done with it, to gather information about how our users contribute to Debian, to gather suggestions and concerns from users, to inform the user base of things happening within the development community and to encourage, most importantly, to encourage users to become more closely involved in Debian. Because of the volume of the results, the user survey results are a fair bit longer. And finally, thanks to Kibbe, Micah and Barry DeFries for helping with the wording of the initial surveys. Now I'll go on to the results of the user survey. So we had about 216 emails. Some of those were mine and most of them were from people responding. Had two translators and overall 98 people responded. So it's not a huge sample size but that's what you get. So most of the people were happy with Debian. Some of them didn't mention this or I wasn't able to infer whether they were happy with Debian from their answer. And there were a few hostile users. Okay, so start off with some locations where Debian is used. So we've got Bird Island on near South Georgia in the South Atlantic near Antarctica on a research station. Another one that was mentioned was Las Tarinas in the Dominican Republic. That's a net cafe. There are a few home users in Sao Paulo, Brazil, Santiago, Chile, Hamilton in Canada. And then there are some university users in near Madrid, in Spain, in Australia and in Estonia. And there's a software company in Italy that's using Debian and the president of a lug in Cape Town, South Africa mentioned using Debian. So some of the projects that these users wanted to advertise tell us about. Firstly, the one terabyte disk-grade machine that backs up 70 notebooks of a company's executives. The next one's the Antarctic Research Station. That was the one that I found most interesting myself. You can get a link off me afterwards and read the email. Another one was a luggers developing a Debian-based robot. Sounds pretty interesting. There's another one, T-Cos is a thin client generation system like LTSP and similar things. Another one was there's a cluster for computer-based ecological monitoring at the City of Newcastle in Australia. There's another cluster. 5,000 CPU cores, one petabyte worth of storage that is used for the search for gravitational waves physics research. Another one that was the Debian blog DebianCuba.cu and a bioinformatics software company wanted to mention the Plone 4 Bio project. So what kind of users is Debian supporting? So three users mentioned that we're supporting their software development. Two were general scientific research. Two were chemistry research. One was talking about computing research. Another home user was doing genealogy research with Debian. Specifically mentioned the GeneWeb package. There was the physics cluster. There was a geoinformatics guy. There were two bioinformatics users. One NecFA, two music studios. Three users mentioned using it in an educational setting. There was one journalist. There was one NGO in Brazil and of the males that I analyzed there were 17 mentioning using Debian in employment or business. So some of the examples of install types. Three embedded, 11 laptops, 28, I mean 11 people mentioned installing on laptops, 28 on desktops, nine workstations, 21 servers, four clusters. These were for ecological monitoring, modeling, the physics research cluster, and computational chemistry cluster and bioinformatics. There was one classroom mentioned running Debian, one data center and one house. The house was controlled with Debian. So what architectures do our users mention? Most of them didn't mention any architecture. Five mentioned i3, d6, four AMD64, one PowerPC and one Spark. Most people are using Edge. A few people specifically mentioned using CID or Leni. There's one old Woody server and a couple of Sarge machines. Most users didn't mention Popcon. One guy didn't know about it and isn't going to install it. Three didn't have it installed and might install it in the future. A couple of them don't have it installed because they don't like reporting information to external organizations. And nine people had it installed. So what do people like about Debian? So at deep package aptitude, a lot of people like the installer, especially the net installer. One person mentioned that Debian was a romantic. They like that it's non-commercial and not corporate. They like the sysadmin tools. They like the community and the support, the IC lists, forums, documentation. One user specifically mentioned they love man pages. People like the philosophy behind Debian. They like our stability, reliability and the quality. Some people like the constantly rolling aspect of testing. People like the arch support. People like that it's the original project. People like that it's an international project. They like the large variety of software. They like that it's officially supported by HP. They like the Debian logo branding. They like our sane attitudes. They like our decision making processes. They like our strict policies. They like the security support for stable. They like software redundancy that you can get several different programs that do the same thing. They like our long release cycles. They also don't like our long release cycles. They like FAI. Some things that people specifically dislike or issues they identified. The daylight saving switches apparently seem to cause a struggle. Flame wars. Some people said it was hard to contribute. Some people said it wasn't easy to find out how to contribute. Some people don't like that we have to comply with software patent stuff. Some people said the BTS was frustrating for new users. As we heard earlier that the Debian webpage needs a visual refresh. Some people don't like the slow releases. Some people mentioned our arguments with upstreams as something they don't like. Another person mentioned they don't like us removing offensive packages like hot babe and similar ones. They don't like that bugs take a long time to deal with to be dealt with. They don't like that because Debian is so large that it's hard to find information about what you're looking for because there is so much of it. They don't like a relative lack of marketing. They don't like that sometimes orphan packages don't get picked up. They don't like irresponsible maintainers going MIA. They don't like over integration. Like being able to remove one piece of software from KDE for example. They don't like the inconsistency in packaging tools like for example Python support and Python central but two tools doing the same thing. And some don't like that it's hard to configure Debian for older machines with less resources. So some wish lists. Revive DWN which has been done pretty much. Keep on going. Don't be adversely affected by Ubuntu. Have a more often updated stable release. Ice Weasel 3 done. More desktop polish. Different release cycles for server and desktop. Better how tos for creating packages. PDF editor. Clone for total commander. Backports. Support for specific enterprise hardware like LSI cards. One guy mentioned they need a basic replacement for a basic sound editor for like Cool Edit or Sound Forge. Some people wanted a better minimal Debian. One guy wanted sysv in it. They want better VM support. And some people want world peace. Some people want to remove the need for a local MTA. Users want backports to be more official. I'm not sure how maybe merge them into FTP master or something somehow. They want the etch-and-a-half project to become more to happen for every release. And they want us to support old stable until stable becomes old stable. So Sarge into Lenny. They want us to be able to skip from Sarge to Lenny. That sort of thing. So what are you what are the respondents used from outside Debian? Debian multimedia.org for all those patented codecs. Cinderella. DVD rip. TT which is a command line time tracker written in Perl. Flock which is a web browser. The Apache Condor project. Plane 3 isn't available in etch. So they were using it from outside Debian. Mod Security which is going to be back in Debian soon I think. Zed which is a text editor that was removed back in the Sarge days I think because it was abandoned upstream. Webmin. Probably shouldn't be using that. AVI Demux and Transcode. Okay. So most people didn't mention DebConf at all. Or one guys is coming to DebConf who's an Argentinian. Francisco Albani if you're here. You're here. Okay. And one guy was friends with several Debian UK developers and supplies them with beer often. So how long have people been using Debian? Okay. 99, 2000 since Potato using SID since 2003. Since Slink was released. Since Potato was released for about a year since Sarge. Since the days of seven Floppy's Netinst which was way before my time. Five years, two years, ten years. Since 1998. Since 2002. Most users didn't mention any contributions to Debian but some did and they contribute documentation. They sell services built with Debian. They run Debian mirrors. They promote Debian. They send patches to Debian. They file bugs. They participate in the forums. They buy CDs from Debian developers. They donate to money. Money to Debian. They donate to upstreams. They use a free software friendly ISP. They wear Debian t-shirts and they serve tea to their guests in Debian mugs. So some barriers to contribution for our users. One guy specifically mentioned that we need an analysis and appraisal of what it takes to become a DD in terms of time. That it's not easy to find out how to help. One guy wanted better mentorship to being a DD. Perhaps something before the NM stage and wanted the process to be shorter. So what does Debian mean to people? Quality software by the people for the people. The best destroyer around. Stability. It just rocks. I love Debian. Okay some quotes. Never sell out to the blue meanies. Speaking of working, I'd better do some. I'm not sure if I've had enough coffee to write properly yet. Wearing Debian feels macho and geeky without getting too slacky. Freedom from the shackles of the Microsoft monster has never been so sweet. Okay now the NM survey had about 31 emails. 26 people responded. Most of them were NMs. A few less people just before the NM stage were a few DMs. There was one recent Debian developer and a couple of application managers. One of the NMs had abandoned the process after one year of waiting for an application manager. Several of them are now DDs. Some of the locations mentioned were Tokyo, Austria, Thailand. Some of the issues they mentioned. Firstly most of these have actually been solved or are in the process of being solved with the new additions to dam and front desk. Okay one guy mentioned that the NM process for non-packages packages isn't as clear as it is for the majority of applicants. A common issue was the lack of time on the part of the new maintainer. One guy mentioned that Debian at least ISC isn't a very welcoming place. A common problem is that Debian developers are just often too busy and there are not enough of them. So it's hard to find a sponsor. Some people found the process really slow especially at the dam stage and there wasn't any feedback during that time. It appeared to me that NMs are generally undervaluing themselves and some of them are not being proactive about the NM process chasing down the application manager when they haven't replied for a month and that sort of thing. Some people said NM was too bureaucratic. Some people found that sponsors aren't up to the task of reviewing specialized packages so they tend to get dropped on the floor. People worried about the size of the dam queue quite a bit. Apparently AMs occasionally disappear. Some people said it was hard to know how long NM will take. A couple of reasons for not joining NM were that at this stage they felt their contributions through sponsorship were the best way to contribute. One guy said he was waiting until he felt ready for it. Some general comments. NM is a big learning process for most people and they found it really helpful. People like being sponsored because they know their work is reviewed and therefore what gets into Debin is actually good. Some people applied too early. Some people were overjoyed at reading their NM report being posted. Some people don't need the new DM process because they've got responsive sponsors and they are in packages teams like Debin Perl and Debin Games. There is a level of pride in being part of Debin and having a Debin account. Some people said that Debin DM wasn't so useful to them because the low number of non-commented packages that are uploaded only once or twice. One person said that not all teams are good places to work. One sponsor said that there are many more uploads to sponsor than time and it's hard to prioritise them so they prioritise by whether they know the person needing sponsoring. One person said that the DM feels like a workaround but still liked it. One of the AMs said that philosophy and procedures checks are a little bit lightweight. I got the impression that the NM quality varies quite a bit between applicant. The AMs both learnt from the process too. Apparently the impact of DM on Debin is as yet unknown. Personally I think that's correct. Ungo mentioned that 70% of his NM process was waiting. Apparently the front desk is quick though and during this waiting the motivation is generally minimised by good sponsors and support from people around them. Some teams, people involved in the games team, the Debin med team, Pearl team, some goals that people had to do GNOME 1 porting and removal. Improved Debin with respect to bioinformatics, packaging squirrel mail plugins, regularly fixed RC bugs, take care of their packages well, help out with sites and math packages, getting NetConf released, working on the herd, joining a core team like Security or the Colonel. Some general wishlist suggestions. Improved the NM bus factor. I think this has improved since survey. Added lots of practical things to the process. What kind of involvement these people have in Debin, they participate in bug squashing parties in packaging teams, moving stuff to main like Alpine, adopting orphan packages, defining policy and tool change for smaller subsystems in areas like the D-language or Yorick, to reintroduce removed packages. So some of the motivations that people have for contributed Debin, so they use Debin and they feel the need to give back to the committee. One guy said that packaging stuff can be addictive. Some people, one guy actually likes to read flame wars and he also wanted to see how a project can work without continuity leadership. So the future of the survey, obviously I'll be posting the final summaries to Debin develop announce and I'm not sure whether I'll do it again. Maybe and if more people are interested in helping then I probably would. After that, after next year I don't think I'll do it again. It's a lot of reading and I'll probably hand it off to the press team or maybe the DPL would like to take it up. Okay and now I'll open it up for discussion. Any questions, comments, anyone? I'm not sure if the Debin team knows this but over the past year the Cacha Federal Bank in Brazil that runs the lottery system is using Debin and over 17,000 lottery machines and to run their lottery system. By bringing to the development in-house instead of using a proprietary out-of-house system vendor they were able to reduce the time of inventing a new lottery game from 10 months to three weeks. So they can suck or remove people? You can talk about anything you want to about using the lotteries and stuff like that but it also helps to fund their social security system and their educational branch. Okay. But you know this is one of the things where I tell people that it's not necessarily the price of the software that makes good software. It's what you can do with it and it's what the value of the software is and the value of that particular software to Cacha Federal was multiple hundreds of millions of dollars. Of course it'd be nice if they contributed that to Debin. Thanks for that. It was very interesting. I mostly agree about the comments you made about the user and the enemy server. This all sounds very true. As if you could you say that anything has happened since the survey that people might have noticed there was something wrong they could immediately or easily fix and that is on the way now or what has already happened? I mentioned each and a half quite a bit. I mentioned the possibility of constantly usable testing and the releasing testing and I mentioned the I think I mentioned the additions to front desk and dam once or twice. Can't really think of anything else at the moment. Well I was thinking of the other way around points raised by users that we could maybe easily fix now not something that was ongoing and appreciated by users. To remind myself of what I talked about there aren't really any that are easy or any that I can think of that are easy how to to fix them. What was this NM bus factor thing you were talking about? That sounded interesting. Interesting it was basically that at the time there was only one person processing accounts and I think that was Elmo at the time that the survey was carried out. It was about the process not about people feeling like being proud of being in NM so that's that would be another bus factor. Was it about that people would feel proud about the bus factor of being in NM or there was the other way around of people doing the process from the back end. Just looking at the current NM figures it actually looks like there's only about four people waiting for dam and the huge stopping point at the moment is getting an AM assigned. So that's actually something that developers can do. You can sign up you can be in AM. You can help work out that backlog that's not relying on one person. There are templates available and I did it in the past and I'm thinking I should probably start again given that's where the problem is. So if you're worried about getting new people in you can make a difference. Definitely. Myself. Okay. I don't know if this is correct for this is a special talk but you were saying what people like and what people don't like about Debian and now this is a particular case in my Linux user group down in Bayer Blanca. Since the beginning of Ubuntu and Kubuntu people started to say okay I can get a special cd and install genome and a special cd and install kd and when I said okay why don't you try Debian just because I love Debian. They told me oh no because if I choose just the desktop they will install me both of them and I don't want both of them. So I said okay go to the console. No why should I have to go to the console. That's that's one that particular thing is why most of them ended up in Ubuntu or in Kubuntu. I just wanted to share it. I don't know if it is the proper place but well since you are all here I guess it's a good time to say it. I am not true but I think the installer installs known by default so you don't get kd long. I have no idea I am always installing using Deep Bootstrap. In fact for your users to install Debian you have three first cds each of which has a different desktop environment. So you can install have one cd for installing genome one cd for installing kd one cd for installing xfsc or you can just scrap it all and do it by yourself. Okay so why doesn't that is being published bearing with green and flashing lights in the main web page. I guess it should be the the way hey start here. Perhaps that's not the idea. But that's the contents of the first cd but do they actually get during installation a screen that lets them choose. I want to restore genome. So they don't see that choice. I'm really not into desktop so I have seen that. I mean you were asking me so I am answering but I mean if you just want to get the Debian default install that it will be known and forget about it. If you want to install a Debian system which has by default a different desktop then it's also there. You get the proper cd to do the installation you want and it's not widely advertised because well well I think I am not into any of those teams that if you if users don't have to do to take a choice it's bothersome to present it to them so just install the default. If users are looking more into I don't like genome okay they can go and search for something else that's it. Wouldn't it be possible to make the choice if you choose desktop environment that there's another question which asks you I want to use GNOME I want to use KDE or something like that and just leave the default to GNOME. Is this a GNOME versus KDE flame war session or is this a talk about the new maintainer process and survey of users? So we had in one of the earlier talks today we were talking it was actually in our DPL's session this morning and the discussion afterwards we're talking about the fact that part of what makes this whole thing work is for each of you who are involved in the project in some way to feel empowered to do something and so if there is something that bothers you about the way the current installation process works or the choices that are available go investigate it and propose a specific change preferably in the form of a source code patch to something along the way that that will carry a whole lot more weight than saying hey you know couldn't we do this couldn't we do that the answer is well yes we could do many different things but the people who are currently working on it are either satisfied with the way it is now or they have some plans that they will work on when they get to it if you don't like it then do something about it that's how this project works. Yeah on this particular from in fact in the in lieu of anybody else with any other points I believe that the where we've got to today is we want to minimize the number of questions that people have asked you on the install so people given the default either CD number one or DVD number one will get known by default just because we need to make a choice it's not necessarily the choice that we you know that we decide we like know more than KDE we just we have to make a choice the there is support already on the on the first dvd for whichever desktop you want you can specify it but it's done by a boot command line there is still plenty of scope for discussion as Bdell said discussion and or changes if people would like the behavior to change anybody else okay moving on to something else um how many people here are AMs hands up and how many of those have actually active and have have been through an M with with new maintainers in the last 12 months cool okay thank you very much to those people please can we give them a round of applause it's it's one of the best things that people can do to get new people into dabion and if anybody else is interested please really sign up it is one of the places where we've got huge amounts of work to do and not enough people and it's a really high-profile job it's a place where you can get to interact with some really enthusiastic fresh people who want to do everything they can the longer we leave people stuck in this queue this is where they lose their motivation and their enthusiasm the more we can we can pick them and steal that enthusiasm the better i'd just like to add that it's also a very interesting way to learn about interesting things in the project because you get into contact with people who are actually doing other things than you are in a very interesting way you get to learn things that you even didn't know existed before and it's a very very nice way to learn the project as a debion developer thank you yeah sorry yeah um on nm stuff firstly one thing is that lots of people do complain about the length of time they're in nm but for a big percentage of those they've also introduced big delays into the system themselves i mean as an am you it's not a tall rare to have three months six months a year go by before the applicant actually gets around to answering some questions and obviously you can hassle them but it's that's up to them if they don't want to do it um the same applicants though well kind of justifiably but would kind of justifiably feel annoyed if you then take six months to get back to them whether they question whether that answers right or not and as as the applicant obviously you want the you want to get a response back quickly but equally there is some give and take there i mean um and the other another thing is that although a lot of applicants in principle would like to have a more hands-on active process of trying out like building new packages and so on rather than answering boring questions that when people have tried this or at least if people try this as a kind of exercise rather than just working on the packages already working on it to ends up generally taking more time and can actually be quite frustrating to work on some some package you're not really interested in or to build artificially some package that's not really going to be used um whereas i mean if if you actually already know the Debian system even though answering the question is maybe a waste of time if you actually have been working on packaging already probably it won't actually take you very long the people who it takes a long time for the ones you haven't who don't actually have the knowledge yet that they will need in the end anyway so you were talking about that's loud you were talking about these delays between uh between caused by the applicants so by the way i'm in nm and my am is he uh mark if he's here great yeah well no he's been fine he has been fine um but anyway you were talking about these delays and it's sometimes he sometimes is the applicant sometimes it's the am's are there any data on the on the latencies like are we recording these roundtrip times effectively yes uh anybody who wants who i guess has access to the maintainer uh mailboxes can tell it's pretty obvious especially because what usually happens is there's an inactive and the main am sends pings or vice versa i mean i've been in both cases depending on what i'm doing so that's usually how you can tell if you really care another thing that you can look at is the last status um it's the last status has the fact that the am has pinged then then you know too what's going on uh so uh so i've been trading mails with mark for the past six months or something um it's when you say the if the last status of the am is pinged is that recorded in the nm cgi's or uh and i guess i what i'm really interested in would be if actually these mails were logged and we could just like see you know over the past 20 nm applicants what even just the date headers were logged date and from just to get those roundtrip times to get a sense of who's who's delaying and how much on which side you'd put them on a web page and unamized or maybe not and then say oh this is you know clearly am's need to this am needs to improve his job or yeah problem though too is the ams i mean ams are often overworked because they're usually the people who we basically take on as many people as we can possibly handle because we know that there's always people waiting um so you can't really use the data to recriminate the am because but the most important thing is to try to identify ams who are not keeping up but as far as the last status thing anytime an am touches any field well actually anybody touches any field in the database it automatically updates when it was last touch and a classic thing most ams try to anyway anytime the senate has been a while back and forth they'll update the little comment field which coincidentally updates the status field so if that's happened then you know that it's not the am that's causing the delay i'd like also to add that um well figuring out which ams are inactive is from the job it's our job so if um if there's something going on and your am isn't replying just mail from this that's what we're for um i mean every am is a person people can get busy can have a problem that's not a serious issue as long as we deal with it and um getting data and putting that online somewhere and somehow incriminating these people isn't really a good thing to do right i can i can appreciate not wanting to incriminate people um i guess even if and i totally understand that everyone's busy including me which is why right now i haven't replied in a month um since i'm here or something but um anyway so but even if you totally anonymize the data and just say for these for this pair of am anonymous am comma nm um if you could show dds hey look the average round trip time between these males is three months uh across the board wouldn't maybe that would encourage more than used to be ams okay i don't think so it seems that there are now much more effective ways for people to participate kind of before they get all the way through nm with sponsorship and team formation and so on is there is there been any kind of study on how that's changed people's willingness to participate willingness to engage just the dynamic of all of this um well most users didn't mention any of that sort of stuff the few nms that that did said that they're um they really enjoyed working in teams because of the support from their fellow developers um yeah there wasn't much information in the in the responses because there were so few about teams seems seems like a seems like a really positive positive thing to keep people engaged i wonder if there's any indication that it's sort of raising people's motivational levels to stay engaged through the through the nm process i can report anecdotally that certainly in a couple of instances i know it has encouraged people to apply to go through the maintainership process which they have seen around them being applied quite quickly and and that everything seems to be moving so yeah i i think that it has made a change having the alternative approaches i think nowadays most people are getting their motivation from being part of some packaging or other team so there's really quite some people who are active in some team and then eventually will apply for nm and then have enough experience to be able to go through and have proven to be enthusiastic about it yeah what i also found from the team survey that i did um admittedly this is completely out of total i've not done thorough analysis of this um but the larger teams with a good number of nms involved as well as the dds those um those nms were all seeming to get through the nm process that much faster um you know they've got ready sponsors around to prepare to upload the work for them they've already got a very good view of how you know day to day debbie and packaging is going without necessarily having to work it all out for themselves from scratch they've got you know a large amount of support around it's probably it well there's no probably about it for me it's clearly the best way for people to get into debbie in these days um to go with the fact of course that you know um the teams are clearly the you know the new way that most of our packaging is being done um everybody wins okay uh we're running out of time so let's cut it short if you would like to read some of the stuff that i've written in the survey results and um ask me any questions see me afterwards and uh thanks for everyone who sent me an email