 Welcome to Neil and Enrico's AM Boff. I'm sorry I haven't got a better introduction today. I'll just hand over now. Hi. I can introduce... ...by becoming the stage guy, so I can just introduce. This Boff is a kind of quick idea we threw up. We were talking about how we do as AM, and things we like, things we don't, and so on. And we had just had the idea that AMs usually work kind of alone, and it would be nice since there's many AMs around EPCOM to just sit down around the table, well, kind of. And just tell each other how we actually do our AM thing, and we can ask questions to Adam. We are lucky enough to have one around here. See, because maybe someone's got better ideas than me on how to do things. Someone found a way to write the AM report in a way that doesn't get you bored to death for a couple of days. Someone found a nice way to go past desk and skills by doing some practical stuff instead of lots of questions. I don't know. I have no idea how other AMs do their thing. I don't have any AM living in Bologna, so we can't hang out at the pub to talk about how we do things. So the idea was to try and do it here. There's no pub, but beer. So, yeah, that's the idea of the both. I don't know if you want to add something. No. Okay. Due to recent being busy with other things, we just didn't have much time to, any time, to actually structure the both a little bit and see how it's possible to facilitate such a discussion. But, so, does something have any, did something come up with a way of doing some part of AM which is, or she is particularly happy with? Like, does anyone thinks he has or she has a good idea that works and he's working? Okay, right. Okay. Mion suggests to start with an introduction. So, how many of us are AM? Quite a few. Quite a few. How many of us became AM recently? What is recently? I have no idea. What, it's a, recently is a subjective perception so that if you consider yourself a recent AM, then... One could define recently being in the, an AM committee, which means has approved an AM and got it through in the last six months. Okay. Anyone is AM and not in an AM committee? Couple. So, I guess I'll start with my own experience and we bring it on. So, I became AM, I know, what was it, a year, a year and a half ago? It was a bit puzzled about the whole thing because it couldn't read AM reports of other people. It was all right. I mean, going to get PHP and task and skill was kind of straightforward. You just start following the templates so that kind of flows by itself. But when you actually get to write the AM report, there's lots of mystical things involved like how to format the email mailbox. And you need to have like a mailbox and a formatted version which is printable with all the irrelevant quotes removed. And you have no idea of what quotes are irrelevant. And you have no access to previous AM reports. So, that was kind of a first little bit of a shock to me. So, I just went on and tried to do my best and turns out it kind of worked. And TBM gave me a couple of suggestions as front desk which I tried to use for the next things. And that was about it. The way I do stuff at the moment, I didn't process many people. I do one at a time. And sometimes I'm inactive as well for lack of time. So, I'm kind of doing my not so good best. But usually the way I do them, I use the templates. And I try to cut away questions that I don't feel they are relevant to what they do. So, if someone is maintaining Python libraries, not sorry, someone is maintaining Python programs, I'm probably not going to ask about library questions, see library linking questions on the ground that he's probably going to get through that when it comes to it. But I still don't know if that is a good idea. And I try to have a look at their packages, what they do and try to give them suggestions about useful tools they have and other parts of them and they may want to get in touch too. That's I think the only two customizations I apply to what I perceive to be a kind of ISO standard AM ship. And I think that's all for me. I don't know if you want to go on. Well, certainly my experience was quite similar to Mdico's. One of the first major mistakes I made with my AM report was attaching the full mail log and putting it on a public web server somewhere. So, that was available for download by anyone and I quickly got told off by Marquis and Ganeff. I have never actually been through the NM process, so I guess this is a kind of a stupid question. What is supposed to be the design goal of NM? I have heard various things from various people. The obvious one is that we want to test to see whether the candidate is going to help improve Debian as an OS. Russ Albury mentioned that the fact that our NM process takes months, maybe years, is a positive thing because it does tend to separate out people who are not truly dedicated because anybody who is not really dedicated is going to last through our process. I'm not sure that was the design goal, but it seems to have worked out that way anyway. So what are we testing for? Would somebody like me actually get through the NM process as it exists today? I am not at all confident that I would be able to, probably too idiosyncratic to pass through the NM process anymore. I'm sorry, I'm rambling. So if you could just answer the thing about what are we testing for, do we recognise that there are different kinds of workers that we need in Debian and does the NM process cater to such variety? I would like to answer that question in my case because what I do with the NM process is that I've had only two applicants, both of them dropped because of lack of time, but still what I do with them is that I consider them already Debian developers in a way, like Debian maintainers. They have their packages, they don't have upload rights, but I try to keep them sponsored as much as possible. I tell them to go mentors and try to get as much information as possible from other Debian developers. What I do in the full NM process is just complete that feeling I have of them. There should be Debian developers, they should know how to upload a package, how to create a package, if they want to be packages of course. I test the rest of the PNP and task and skills by trying to deviate as much as possible from the templates. So I take the templates, I modify some things, I send them and then I pick things and then I go through them a bit further. The first thing you said was that you consider them Debian developers because they have packages and they are maintaining them. Which begs the question is a Debian developer just a glorified package or is being a Debian developer, does it go beyond just packaging up some piece of software that you happen to like and use? I personally think that if all we are doing is testing packages, we are putting way too much effort in something which does not actually require this much. We can probably automate the process. Creating good packages is something that is a testable skill. We don't need 18 months of NM to just create people who can package them in packages. If I may, I wouldn't want this both to become a let's reform NM because we are going to have it this afternoon. Although the question of what do we want from each of us, I want this to be more of an experience sharing between AM. Although the question is what do we expect, what do we look for from a candidate is I think a very good one. Everyone's probably got different things. Jesus considers them to be Debian developers already and tries to fill in the last steps. I personally would like to get myself convinced that this person could be a useful workmate basically. I understand the frustration with the long thing. I went through in a week and I guess if it had taken more for me, I wouldn't have gotten through either. I try to keep it short for my candidates as well. Normally my candidates went through fairly quickly except this last one which is just waiting three months to reply my mails. He says it's very busy so that's not a big deal. Other people, what do you look for? I think handling bugs is a big part of it and that really can't be done in a very short period of time unless they've already done it before they apply and how much commitment they really have to the project which you can come in really enthusiastic on a project and then give up within two or three months. I took ten months to go through NM and that was about three years ago. I'm now working on my first application. Others? I just wanted to complete a bit what I said before. When I said that I was trying to deviate as much as possible for the templates I meant that I tried to look for other skills in the person. I read his emails, I read his answers to mainly lists. I see how he collaborates with other people if he's a flamer or if he's a total ass trying to make others feel bad as I've seen some people doing it. I don't believe that person should be let just inside a project where you're supposed to work in teams and you're supposed to collaborate with other people just immediately. I mean that person should learn to collaborate and to work with others without engaging in... Sorry? I don't know how good a collaborator I am. Right, but even you have the ability to work with other people when it comes to... Not demonstrably? We're working together right now, to some extent. Part of the thing that has to be able to, even if you prefer working alone, is ability to communicate with other people when you have an issue to talk with other developers. One of the things that happens is when people can't do that, then they get in a situation where they're ostracising themselves from the project. Maybe some people prefer to be ostracised, but in part it's something that maybe we should be looking for a little bit more closely as the interaction between developers. I guess working together in this case means like reporting a bug report decently or reacting to a bug report without being offended. I mean that sort of basic things that makes it actually work, rather than I don't think anyone can get rejected from NM for not answering a bug report in a cheerful way. So, luckily enough. There was someone, I've seen the microphone rolling around. No, okay, it was rolling to the microphone holder. So we can go on, I guess. No, Maree? For me, I guess, although it's not necessarily what you're thinking about on a day-to-day level, the point of AM really is to find out whether the person is appropriate to be a member of the community of Debian. People are saying about working together. Again, it's not just on whether you'd like to co-maintain or become part of a team, but everyone in Debian does need to be part of that community in the way that they interact with other developers or with users. But another point quickly. I think a lot of the people who spend a long time in AM, sorry, in NM, I think often it's because the AMs are generally very reluctant to reject people. So some people just stay in NM until they give up. That's the most common way for people to leave NM, that they are in it for a long time and then they give up. It's very rare for AMs to reject people before that. That also sucks. You get into some kind of contact with someone and you see that he's putting some effort in it. At some point you don't feel like you want to turn them down, which is, I don't know, maybe a problem that we're having now. Anyone's got experience about it? I actually wanted to ask you a question of every AM. Could somebody who has done the task-based tasks, where instead of asking questions, you assign tasks to your NM, like talk through how that worked and how you, I mean, what I'm most interested in is how you figured out parity between the tasks and task and skills. One of the things that's scary for new AMs coming in and abandoning the templates is that they don't necessarily know a priority when they've achieved parity between the task questions that they're asking and the policy questions that they're asking and the tasks that they're assigning. Anybody who's done an AM who's done the task-based, so deviating from templates, I'd love to hear from you. Now or later, whatever, it doesn't matter. Has anyone here actually, instead of using the templates, used the task-based system? I haven't done the task-based check with any AM, but I've got some feedback from the front desk about it who did it with at least one applicant. And he said it was much, much more work than the usual template-based checks and used much longer time to get this stuff done. So it wouldn't help with the time some people need to go through an AM. My fear with the task-based thing is that you end up doing something that the person is not interested in doing. I don't like the task and skill question like pick up a random package without a man page and write a man page, because it may happen that there's no package without a man page that this AM uses. And so you end up forcing someone to write a man page for something he doesn't totally care about and he doesn't know which could be compensated by just going on and looking at a man page this person has written for a package which already exists. So that's my main fear. Well, that is a problem I would have with task-based things. Although you may agree on a task together instead of assigning it as like, do you see we can fix an RC bug together, these sort of things. Although the other fear I would have is how do I format the AM report? Because that would generate lots of emails going back and forth and maybe IRC logs. And so it's very easy. We almost have a recipe although boring for creating AM reports with the normal tasking skills. You remove some of the quotes and you're done. Well, you remove quotes, format, well, but it's kind of straightforward. It's just boring. But when you have lots of emails and you have to format them into a printable version, then you lose threading. That's like another thing I wouldn't know how to do. I've been tempted to say let's fix bug together. I mean, it happened to me that I've been post, I've been attaching an IRC log which DNM sent me as a signed email because it just felt like that was a nice way of doing it. Because we were just chatting for IRC and I asked them about how to do something in WN and they knew it. So it was like, well, we were working together but I also happened to be your AM and you just answered me a technical question I didn't know. I guess I should put that in the report. But, yeah, that's kind of cumbersome, I don't know. Morhe? I was trying the task-wise system with one of my applicants and it utterly failed. Mostly because I picked the wrong package. It was a package I had just often before in the MIO process and some other person just wanted to adopt it. So I ended up blocking the adoption process. So maybe it was just because I picked the wrong package. But at that point I noticed that it's very much more work. You have to look through the bug reports yourself. You have to see which bug tracking system is upstream using how is the applicant forwarding stuff. And then probably working on the packages also requires considerable action on your part. Is this certainly something I would like to do, but next time I would have to prepare stuff much better before? Well, I think we should consider every NM's process. Even when we use the templates it is still task-based. I mean, the most important thing for me is how they're doing their ordinary work in Dublin. So if they're a package, how are they maintaining their packages? And the templates you can think of really as just their supplementary. That it's just a way to fill in information that you're not getting, that you're not seeing easily. So it's just a shortcut because rather than waiting for them to demonstrate every single aspect that you want to ask them about. But it still seems to me that the questions are not the main thing. The main thing is how are they working in Debian in the normal way. Which also brings to the idea that I hear more and more often that people shouldn't go through NM without an existing record of activity inside Debian. That may be already a third line of filtering, but I see you. No, no, they shouldn't go through NM without a previous record of activity in Debian. Because that would also make it easy to test. Actually, I wonder if it would be possible to just do AM process by googling the person, seeing mailing list archives, and just asking the person to sign the piece of mail which says I abide to the Debian social rules. No, sorry, I've been talking about the community guidelines too much. To abide to the Debian social contract and the FSG and that signed mail we ask. And then as an AM I just say well I've been tracking what this person has done in the last year with Debian. And I think he's just right and you add links to relevant stuff. I don't know if that would be any useful. It would probably work for some people that are very active in mailing lists. There are multiple people that are very active in their own packages, getting them done in a very good way, maintaining them, are very active with the users and stuff if they get questions, but are not very visible on Google or something. So it only works for a small part of the applicants I think. Right, but if instead of Google I just ask them to links to mailing list archives or how they close bugs and I don't know. If I can produce a little bit of proof or links for that, that could just replace task and skills? Yeah, it may replace some parts of task and skills and also PNP as you can find some, how they behave and what they know and that stuff. It won't replace all of it, but parts can for some people certainly be replaced with that. Which actually prompts an interesting thing because I also feel that most of the things I ask, I don't necessarily want to know them, but I feel like I have to or front desk won't be happy with the report or the dumb won't be happy with the report. Like for example when I ask a question, when should I consider myself happy with the reply? I would be happy if a person just replies with a link to where is the piece of information because that means that the person can actually find out. Yes, that is one of the intentions of the templates. You don't need to see it as a policy completely. It is good if you know where the stuff is written down and you also give a big set of links in one of the first templates. So pure policy, copy and pastel doesn't help anything. There are some questions where you need to write your own stuff, but for most part you can point there, you can point there and show that you know where you find it. There are also that many questions in the templates that you can consider. Some of them are very optional like the famous linker questions, which are nice to know if you ever want to go and package a library, but it's also good for most applicants if they know where they can find the information. So some short sentences about that is also okay. Thanks. That was a good piece of information to have. Mina? So maybe it would help to create an AM report template. So we know what the thumb or what we should expect from an applicant to know. So the currently existing templates are just optional. So when I know my applicant knows how to package libraries, I can just drop all of task and skills, but then I need something to write in the report. So at the moment, what I probably can't just say, yeah, this guy maintains a whole bunch of libraries, does his best, has gone through several transitions and whatever and whatnot. And yeah, maybe we need some guidelines for the AM report to write to. Guidelines are hard to write out, but we can try to get some of those. It's hard to write guidelines that fit all of the applicants like we had, I think what Albury was said was done some very intense task and skills set with his AM, going through a fucking huge transition, which was only for one library, but got multiple other libraries sucked in getting those from unstable to testing. Like 50 or 60 males alone in the AM report. And for people that manage such a transition, it's very easy to skip other parts of the task and skills, but because based on that you can see that he knows what he is doing and that he knows the stuff. I think we've got another question in front there. Hi, my name is Barry Hawkins. I'm actually a new maintainer in the queue. Andy Bart is my application manager. And just listening to you guys and then having listened to the NM process video last year, it seems like a manogis question is one of the most pertinent, because until the destination is defined, you won't really know when you've gotten there. And until it's a little more clearly defined what you want out of the process, it's kind of hard to say, well, we're getting closer. To give you just a very short case study, I applied, I was told, hey, go ahead and apply. You've been maintaining packages for months. I waited like seven months and then got a message where I was initially turned down, actually. I work as part of the Debbie and Java packaging project. So my name is not in the maintainer field of any of the seven packages that I've been maintaining for a year at that point. So, you know, we kind of worked that out. And the advice that was given to me as I started through PNP one was whatever you do, don't screw up because then they might tell you you're not ready yet and then you got to go all the way back through. So in PNP phase one, I probably have 30 to 40 hours invested in answering those questions that consist of about 10 to 12 pages. There are actually two weekends out of my life where I told my wife, I'm not getting anything done this weekend because I'm going to sit down and go through these things. And quite honestly, the process is so obscure. I don't know if I may be rejected or if I'm boring the hell out of Andy Bart. I don't know, but it's literally it's such an obscure process and I really don't think the destination is defined and so that it's kind of tough to know how we're going to get there. And I think the last thing that's going to facilitate the process is, well given that we don't really have the destination that well defined, let's actually make the process more complex by having an organic process where you kind of try to adapt to the individual situation. I totally see the merits of adapting it because certainly as a Java packager, not spending a lot of time in the shared C libraries, you know, the stage of life. Anyway, just an observation as one who's in the meat grinder currently. It's basically a more like a question. If I have an applicant which I know he is good at packaging and I see it everywhere on the mailing list and so on, can I just reference that or do I have to do a task and skills check for each package anyway? You should look at the packages if you find something obvious, but if he is doing a good packaging job since a long time, you can just write it down there and list some examples and stuff. You don't need to do intense checks then. I see our time is running up so we could get to the point where we have some solutions. There was this proposition some time ago that we introduced some system where we have centrally stored mailboxes for each AM, which would be optional. Would any AM, maybe an AM object to the full AM report being published to all other implication managers? Because it seems like no one knows what the others are doing because no one has seen another report except for the front desk and the account manager. So I would like maybe to open the AM mailboxes, the front desk communication, I don't know which part of it to maybe all of Debian, all of the application managers. So are there any objections at that point? I personally wouldn't mind, but I would suggest to make it optional, so ask the NAM or the AM whether they want to, because there might be NAMs who don't want to have it published, there might be AMs who don't. It would be an internal publication just for the other AMs. So personally wouldn't like if it gets public for whole Debian Well, the templates itself are public, but I don't think the whole report should be completely public. The NAM committee gets some reports from every now and then when every dumb rejects a complete... The NAM committee only contains experienced AMs, not the beginners. Yes, of course. Also the front desk is reading the reports of every new application manager the first one or two completely, and giving them tips, hints and free marks wherever he has something not done completely like it could be done or where he missed stuff that you should do. Maybe those things could be done to the NAM list, AM maintainer, to the application managers. So more people see what could be enhanced. I should now have a question for the dumb. Suppose I try, like now I'm a bit lightened up from this talk and I may be less of a bureaucratic nuts. I would fear less that if I don't fill in every single bit, the application is rejected. So maybe I risk exceeding the other way round, like making it too quick and the applicant is not very well assessed. If you are not happy about the AM report, do you get back to me or to the candidate? It depends on what is missing or wrong in the report. If it's clearly an AM thing that he completely missed it and you can see that the applicant itself has the knowledge or could have it if the report would be done in the right thing, then we should go to the application manager and educate him to avoid those mistakes in the future. But at the most times I've seen a bad report until now. It wasn't only the application manager, it was also the applicant having some... he missed some things over the whole report and didn't look like he involved. Okay, because my fear was like suppose that I now start experimenting and I make wrong choices, but then the applicant maybe spends some time in the queue and then the dam gets back to him saying like, hey, this is missing, tell me about this, this and that, and that's kind of like he's paying for my mistake. At the moment, front desk is usually set fast with such questions, so they are cleared by front desk shortly after the AM sensor report. Another question I would have is to... Is there some better means to communicate with each other? Is the WNU list enough or do we want something more private? Maybe something like the comedy list, but for all AMs to talk to each other? We could try that, it's no problem. New man could be for public and if it's something AM only, like if you have a really bad applicant and want to talk about it, what to do with it, we could certainly set up an alias on Merkle, which is an AM debion org, to only communicate with all application managers which are active at the moment. That would also be a good place to publish the reports that are both by applicant and application manager agreed to mail to other application managers. While there might be a need for a private channel, I would say that we should have a specific requirement for hiding communication before we do so because as far as possible, we should be an open project. If we do have an applicant which is problematic, maybe we could still communicate about such an applicant in a non-secret manner as long as we remain polite in our emails. That's a very good point. Remaining polite in that with debion's flambos. Yeah, that's of course a good point not to hide anything, but I don't think it helps in this case because if people do not write mail because they have to be polite and don't want to, that's not a good thing in this case. We want to be open to the other application managers in that case. So I think a private mailing list for AMs would certainly help. We have the open mailing list there, so we only have to take care that we don't hide public information. There's also the case that some points about applicants you may have shouldn't be that public because they can hurt his reputation because most of our lists are archived on Google and if you search for his name, I don't think everyone wants everything but other things about him in a process like getting into debion in the public archives. Yeah, I guess one scenario here is avoiding, because in NM you can just say I don't know this, but you may not want it published on Google since maybe five years afterwards people will find that and have a different idea about your skills. That's probably the most important scenario we're trying to get out from where there may be a solution which fixes both, like be an open project and not publish long-standing incorrect assessments about people. I would see, for example, another scenario that I would like to have a private channel if I receive a bad report about my applicant and I want to verify that that bad report is in fact true. I would like, for example, to communicate with the German group because my applicants have been German, so I want to get some feedback if they know him, if they know the community that he works with so that I don't have to publish on Google that I'm trying to look for someone who stands for this guy. I don't think there is a need to establish a bad beginning for a guy that is just trying to get into Debian. Although there's Debian Private, I guess, could be used for that. Private Nail, Debian Private. It certainly can't hurt to just try it and see what comes out if he has such a private alias for all application managers. It can only fail and then you can go to the old stuff back and if some application managers also have questions they want to have answered in a more timely manner and not only by mail, they can also go on IRC and ask there. We have a Debian Newman channel on both networks that are commonly used, so people have multiple sources of... That I didn't know. No, you know it. Debian Nail? Debian Newman. Okay, no, I didn't actually. Mainly on OFTC. Okay, that's good to know. Well, I'm not an AM, but I'm trying to encourage people to become Debian developers if they are doing great work. So I managed to get one maintainer in, the second one did give up after seven or six, eight months and the second is now in the queue, the third one is in the queue for I think 13 or 14 months. For me as a non-AM it's nearly impossible to get some feedback or some status about the process. Well, I can ask the actual AM, but it's nearly impossible for me to fast-track that kind of work or the person. Well, if I propose somebody, I have to, in the meantime, have to do a lot of NMUs for him, do a lot of work for him. The only thing I did hear from the AM was, well, if you do not want to do that work anymore, I can do it as well instead of fast-tracking the process. That's a bit, well, strange. I think that also depends a bit on the AM those people have, what reactions you get back. Some AMs may just answer that the applicant is waiting for mail from him or the other way around depending on who is busy. Well, what I would like is to have the chance for a normal Debian developer to be, that you can see the process of NM if you are interested in NM, that you are able to follow the discussions. Following the whole discussions is a bit lot because you have a lot of emails exchanged between the AM and the new maintainer. I don't think you want to get a copy of all of that which happens there. It's the most commonly used part for information how far one is, is really the NM Debian oxide which is divided into the usual four steps, ID, PPTS and goddess account. It's hard to get more details into that side because where in the process are you setting more points to give information where the guy is, half in PP, half in T and S or whatever based on the way the AM does its work with it. Okay, we are very quickly running out of time here. I just wanted to quickly say something about what I meant before. My point is not that we need a private mailing list. My point is that I would like to know what I have to do if I want to figure out about a person without raising a flag on him. I want to know what we have to do to be able to share information between AMs if we set a subversion repository that I can check out other people's reports to go through them and to learn about them. I want to know what I have to do when I have a person that is being not communicative enough and I want to figure out more about him. All the processes that should be more clear and should be defined instead of trying to figure out what I have to do when I have a problem with a person. That's good. That's the last intervention and then we go to close. Since I brought this up, maybe I should clarify this. We already have a private communication channel. I'm not sure we need to add to it. Most of the things that we are talking about ought to be exceptions. These ought not to be the regular modus operandi. This is not the way we conduct business. I understand that there are exceptional cases in which we need a private channel, a back channel for communication. I think even if you have to go to Debian Pride and it goes beyond the AMs, that's not a disaster in the few cases where you do need to employ a private back channel. I would stress that Debian private would be considered an option if Debian private or maybe we should have a strong reason why we need yet another private back channel. I just wanted to interject one thing really, really quick about the status. What you can always do as AMs is use the comment form in the database. One of the advantages using the comment form is that it updates the time in which you last saw the report. If email you send PNP1, you can say sent, you can say PNP1 received. So you can fill that in and that will give everybody else information on what's going on. Thank you for everyone. I got the impression, I'm quite happy about the both. I got the impression that most of the some of the cumbersomeness of the AM process actually sounds like the bureaucratic effect that happens when you have to fill up the form alone and you fear of getting it wrong so you get more and more anal and precise and pedantic and obsessed about how to fill it in and I myself feel like it was fairly useful to actually share a bit of the experience and ask some questions and see how things could be customized or how much can I apply my judgment. So I look forward to more of this discussion happening outside of this both and we knew, we know now, well I know now I probably was the only one that didn't know that we also have an IRC channel about it. They have a new main on OFTC and later on in the day if I'm not mistaken there will be another both about reform the AM process. I hope the content from this both Mayone is running it. I hope content from this both could be also useful context for then running this afternoon's both. Thanks for everyone for attending here and now is the ex, if I'm not mistaken, ex, well, ex of community well. Now Natty will present the next both.