 We did this last year, and it was a smaller room with more people in it, which in itself sends a signal to me. Now let's see how we get on. I'll assume that nothing else is going to happen there. Last year there was no camera. This year there is a camera, but I understand that it's not going to be streamed live, so however there will be a record of the discussion. I think one of the best things about last year is that we made it very clear that it's an engineering discussion. So we're aiming just to explore as engineers what the state of collaboration around automotive software is. I think that makes it unusual versus some other panels that you might see in other places because there's no marketing. I've been told I'm under strict instructions not to talk about the politics, so I won't do that. We're just going to talk about the technicality of what's actually going on in terms of how people are working on software for automotive systems. The result of that is that all of the gentlemen here are actual engineers, they write code, they've been involved in real systems. They're not here to sell product or present companies or anything like that, we're just going to talk about the realities of what's going on. I think without any further ado, I'll briefly introduce myself and then ask each of the panel members just to give you a sense of who they are and why they think they've been gathered here. I'm Paul Sherwood, I'm the CEO of Code Think, I think I'm one of the few CEOs that actually writes code. I know there are lots of those in Silicon Valley, but I'm one of the weirdos, so I talk as well as writing code. Probably that's why I'm on this panel. Philippe, please. Hi everyone. My name is Philippe Coval, and I'm working from Samsung UKA. I'm part of the open source group of Tizen of Samsung. I used to work on Tizen and also on IoT projects. I have a little experience about Tizen IVI because I worked before on it. Also, on AGL and GeneVie, this is a two projects we have, maybe Leon can explain one of them. Thank you Philippe. My name is Leon O'Neville and I'm a senior software engineer at Kansuka Group. I'm also an open source enthusiast and thanks to my work, I have the opportunity to contribute both to Automotive Grade Linux and GeneVie Development platform. I'm currently working on the integration of the SOTA project and Austria in Automotive Grade Linux. This is a work funded by ADS Advanced Telematic Systems. It's a really exciting project that Arthur can mention more about it. Sure. My name is Arthur Taylor. I'm CTO and co-founder of a company called Advanced Telematic Systems. We develop software for automotive companies and their suppliers. I'm a CTO who's regularly asked to stop writing code. We have been heavily involved in AGL and GeneVie for a long time now since 2013. Sometimes working on user management, more recently working on software updates, which is how we come to collaborate with Leon and a couple of the other guys here, especially on the AGL reference platform and GDP. Hello, I'm Stefan Disto. I'm a CTO at IOT.BZH, which is a consulting company, which is a kind of reboot of the team working on Tizen for three years in Brittany in France. We started by working for Renaissance and more specifically, even if Renaissance is funding us and is a member of GeneVie and at GL, our team is mostly dedicated to AGL. We are also working and looking in detail on GeneVie to see how we can find some conversions on many domains. Software update is one domain, for example. We also work with Leon and Arthur for that. IOT is also another domain where we can share a lot of things between projects. My role in IOT.BZH is mostly re-engineered, so I'm writing code, but as a CTO, that's it. Marvellous. I can talk whenever I feel like it. Another thing I think we got right last year was we actually invited members of the audience to speak. That was both questions and comments. I've got a few of the key points from last year that I intend to ask later in this panel to see whether we've done better or worse and whether the issues are the same or have changed. Basically, I'm going to ask broadly the same questions this year that I asked last year. We have some continuity because Arthur was on the panel last year, so he can hopefully keep us sane and remind us of what was said. But the other three guys are all new. I've not actually met them before. One of the key points I would make is that there's four different companies here, but they have, I think, all collaborated with each other over this last year. So you've probably got a better interrelationship set. Yeah, it's on point, yeah. Okay, fantastic. So if I could start with you, Philippe, one of the key questions is just if you could maybe give me some examples of the kind of highs and lows of your experience collaborating in automotive. What's been best and what's been worse? Okay. I think the good thing is that I'm not an automotive engineer and just discovering this with Tyson before and I had no special experiences when I was working from the software side, not especially in relationship with an automotive maker. So I think the good thing is that they are adopting development from open source projects, Linux-based projects. So somehow if they are interested in Linux and the related workflows, so I think those workflows were approved before by other projects and if they are following something that works, that's the good thing. And the bad thing, I would say, but this is just my personal appreciation, I don't have experience. Those industries have specific workflows on long term projects and these don't fit with live patching code like we do in open source. So probably there is some unalignment between those two methods and somebody has to sort it out. So to say this in other words, making products and making software platform is not the same thing. Yep, I think that's a very fair point. Okay, thank you for that. Leon, same question to you. Highs and lows. Well, let's start with highs. It's really fantastic that both AGL and Geneva Development Platform are open source projects in the full meaning of this. So even if you are a software developer without experiencing the automotive industry but you have experience with open source software, you know what is GitHub, you know what is Garrett, you can start contributing right away. So this is fantastic. You can very easily fix a bug or add a new feature and get it merged into the upstream. I would like to highlight the work that we have been doing on platforms such as Raspberry Pi which is very popular among the community and we shared a lot of knowledge between the two platforms to get it working there because there are a lot of similarities between the platforms. From the downsides, well, I would say the same as Philip. Platform and product is different things. It would be great if you have more products with these platforms. Yep, okay, that's fair. Okay, Arthur, same question to you please. Well, I think we've had an exceptionally happy year contributing to open source collaboratively for automotive. At the start of this year with Leon's Help, we got our software into AGL and GDP and that was very straightforward and smooth and we were very happy with support from Code Think and others. So that's been really, really good. We've been delighted by how willing and interested that people are to hear about open source solutions to critical problems. Sometimes a couple of years ago we would get pushback on why are you doing this in open source. These days when we go to customers, even if we're talking to engineers and they understand that there might be problems with legal or further up in the business, there's really, I feel like a broader acceptance that open source for non-differentiating parts of the stack is a really good way to go and we felt really supported to bring some of that software deeply integrated into these platforms so that's been really great. In terms of Lowe's, I'm not supposed to talk about the politics so I won't but there are, you know, the frustrating part is always when you do work and you want to share it and it's not accepted or recognised, that's always difficult but I think that's just the nature of collaboration sometimes so we've been very happy with the collaborators that we've found and when the collaboration has gone smoothly. Right, okay, that's fair. Always collaboration is complex isn't it because it depends on the people. I think some groups of people get on better than others and it's always the way. Yeah, you agree. I would say that collaboration is a great point between many experts in different domains because I'm also coming as Philip from the open source side in fact more than the automotive, I'm more in the embedded world than specifically automotive and we can see that we need experts in all domains coming from open source side, Linox gun or whatever and also we need automotive experts who will give them their feedback. They are quite open to give their feedback and so there is definitely some projects like AGL and Geneva where we have many things to share between engineers coming from different worlds. I would say that's the good point. The bad point would be that we could go further I think. The community around automotive is mostly the members in AGL and Geneva. There are car manufacturers here once but you won't see many students for example. So I mean this is a kind of open source community but not in the sense we call a community usually I mean in the open source world. So I would like to see more young people being interested in such projects and give them some platforms, some cheap platforms so they can experiment, they can contribute, they can develop applications etc. So that's something we will try to develop next year. I would say that in the good points I think that many OEMs now have taken in account what we tried to push last year basically the security and automotive in connected cars. That was something we, at IOT.BZDH, we insisted a lot on that and saying that we should build security first inside the system that should be shared by all OEMs and tier ones participating in the project. And finally, I don't know if we won but finally we have some kind of security in a GL so I would like to see the same, I mean maybe in different ways because there is no signal solution but I would like to see the same in Geneva for example to see security models that we can compare, we can improve. Having multiple projects is a chance, it's not if I go on the political field. We're not on politics here. No I know but I would say that it's a chance to have multiple projects because we have things to share. Even if there are differences, I think that there are many things in common. I mean we are both based on Yocto, both based on Linux, both based on many software components. We have the example of software update where probably advanced telematics wants to have a single component for both projects so just one example but we can find many. All right, now I'm going to throw this to the audience. Does anyone care to comment on highs and lows from their point of view? Anyone got anything great to say about automotive collaboration or something they want to highlight as a problem? Don't be shy. Or be shy, that's fine. It's a free world. No one? Okay well, oh yes sir, please. Who has the microphone? Thank you very much. Okay so I worked for a company on the automotive environment of one year ago for a few months. So I agree with them in the sense that sometimes the community is not open enough it seems for me. It seems like a small community among automakers but it's difficult from people outside of these companies to really get into mailing list, get into code, get into, I don't know, it's a feeling I had. Just wanted to share this. Okay, that's a fair point I think. And while the mic's passing I would say that I've been working within an automotive company that is very positive about open source but it has some basic IT issues. It's hard for their people to access IRC, it's hard for their people to be on mailing lists and those kinds of things. So there are still some relatively low technical barriers that are getting in the way as well as the historical culture. Yes sir. I hope my comment is not too much on the politics side because I see there's a lot of collaboration that's very nice to also on researching and trying how to make use of new technologies, many of those on coming on open source on all levels, on the devices on the car, on the cloud applications whatever. But this thing will not scale is my opinion because in the end the complexity is very high, that was your starter and it will not work that every car vendor and I don't know whoever will develop the software by its own, the same piece of software, the same functionality and also like the car vendors work today they rely a lot on to collaborate with other companies to deliver this stuff, the software and also the communication on this collaboration gets more complex and this actually requires to get standardisation on the functional level and that's something where I have not seen any progress and maybe you have from the software side but that would be my downside so that I'm going to take that and put it directly to the panel so if I understand you correctly you're saying that you fear that the current model does not scale and that you think we need to have more standardisation in the open is that correct? I can try to take this one from the software side we are somehow aligned because we are using Stefan said this, we are using the same software component with different variation but it's roughly the same and on top of it we have the value specific from one project to another and this is where we are not speaking the same language so in terms of security in terms of resource management and so on so maybe there is not enough visible progress but there is one solution I can promote here this is just about the definition of the things what is actually a speed what is a door lock what is everything you have in a car it does exist but there is no common name to describe it so first I think to establish communication and it also applies with human beings when you're meeting someone speaking another language you have to quickly find the level of understanding of the other and the vocabulary to have an estimation of the vocabulary and one you will find the reduced dictionary you can start by talking but yes you can also when, where, how and by just simple words you can start communicating and I think it's applied between human beings and it could apply with complex system like a car so maybe as you said there is a lot of complexity but one approach as engineer is try to divide problem into the smallest part and try to fix them and defining what is the resource I'm talking about resource but I should say the simple things are viable and once we are aligned on the name of it we can try to put things together because my previous experience in Taisen for instance which is a very complex system there is a lot of specificities but not cannot say there is a lot of relationship between each component so that's one of the reason IGL wanted to have a different approach on this project so to say this if we have very small reduced definition of what are things and how to interact with them I think we'll make a small but important address and getting it into smaller components I would just say my own view is that the traditional methods don't scale either so I mean I did a graph a few years ago about the increase of software in cars and basically it's just this huge hockey stick so we went from a million lines of code in a car in 2010 to about 100 million lines now and it's going off the scale because we're going to do self-driving cars I know from personal experience that the proprietary methods that people are using don't scale to those levels either so at least there are some very large open source projects there are projects with thousands of engineers that are very effective at cutting code but in a way it feels like the automotive collaboration is still early isn't it we're still trying to establish the body of work and get even the vocabulary right for some of this stuff but I would gently say that I don't think that's a magic solution in private either the scaling problem for cars is a significant one anyway we now have just much bigger more complex projects to do I'm going to try to move on because we're about half an hour ago so let's crack on with can I ask I'm going to move on Leon what do you see are the hot technical topics from an engineering perspective now versus a year ago what has arrived that we need to think about I think this is an easy question we had like a dozen of talks about software over-the-air updates I'm lucky to be working on integration of one of these projects in the AGL engineering development platform but my experience from this conference is that a lot of people are interested in atomic updates of embedded devices not only in the automotive industry but in general so for me this is definitely the hot topic right now yep, fair enough that you're bound to say the same so I'm going to skip over that unless you've got something different to say what I was going to say is obviously it's good validation for what we're doing that it is such a big topic right now obviously what I get asked in my position is what's next and for sure what we see next is as you said these advanced driver assistance systems sensor fusion all the software at the rest of that hockey stick because just working out how to deliver the software is only the very first small step to having these billions of lines of codes in vehicles yep, let's hope it's not billion Stefan hot topic yeah, I would say that if we look at the cooperation between common manufacturers T1s et cetera on projects I would say to answer you I would say that it could scale because the goal is not to create a full system because you need to keep some differentiator for the car manufacturer in fact so probably the goal would be to create a best system which would represent something like 60% of the platform and let's say you let 40% to the car manufacturer and T1s to express themselves but the point is that you want some base services to be in this base platform so Linux kernel, upgrades security model application framework all those kind of stuff should be in the core system and that's our daily work in fact we all know that at some point we'll get a navigation system it's obvious but that's not the point that's why when we show some demos the point is not to show something very sexy because we don't care really about it we want to create a system which is robust and which can last for 10 years or 20 years which can be upgraded etc so that's a real challenge and everyone is committed to working on that I think so I would say its scale is pretty well currently again to the audience does anyone want to comment on hot topics or things that we see from an engineering point of view that need to be front and center in our minds maybe I have one thing have you heard about adaptive autosar or autosar's adaptive platform yes I have yes I'm working for a company and we are starting with the specification there is a autosar consortium and they start with the demonstrator with Yokto language about the build system about your packages then the linux kernel and then some standardize communication middleware that should run on top of this platform and they are working on it they are specifying and they are trying to get whatever is useful and to have a common language between OEMs so how you develop your software how you package it how you make that work is being done in the open then the specification and the collaboration is in the open or do you have to be an autosar member I think the SVN is not in the open SVN did you really say SVN it's 2016 what the fuck today I got a message we got our own bit bucket oh fantastic you just keep right down that line ok I'm sorry thank you for raising that gentlemen to your left so I think the next big thing is really the over the air update but I mentioned it earlier it's not a topic of updating one device but we are updating a network a complex network in a car and the topic of rollback mechanisms and security is the next big thing and I think there is no common solution until now it could be a topic for standardisation but I think the OEMs and tier 1s don't want to do it in a standardisation way because they want to be unique well I take your point but I would just say in general I think the OEMs have been slow to understand the scale of the problem they face and they'll standardise as rapidly as they see a solution because it's hurting some of them now but they are also burnt by autosar I think oh yeah fair point I think there was one more Agustin? It's not a technical topic but it affects the root of the technical collaboration which is the licence and it's something that is getting worse so the fact that as an industry the OEMs has clearly said that they do not want gplv3 it's making things hard but it will make them even harder and it's one of the first times I see an industry for reading and gpl compatible licence so I think that's quite unique The problem is probably wider than simply on the automotive market because for any embedded product as soon as you want to create a product you'll hit this problem so yeah this is something we have to address we already had the problem for Tyson we know how to apply some kind of work around but definitely this is some extra work which obviously we could try to avoid maybe one odd pointer just something I want to raise because it's quite fresh quite fresh it's this year I think it's the fact that OEMs will want more and more hypervisors for their systems and putting something like a mix of GDP or HGL plus Android plus QNX something like that with three or four displays is something that they are thinking about so at some point again this is not that easy because if you know a little bit about hypervision if you want performance it deserves a lot of work on the driver's side but that's good, that's more engineering to do yeah so we still have a huge work to be done so it's good news for us but that's definitely a hot point I traced this year we didn't hear about that before and I will just give a brief plug for Jailhouse Siemens have done a lot of work on this kind of tiny hypervisor specifically for that kind of use case and I think that's a very exciting technology briefly moving on to the negatives that's the positive things that are exciting pain points, engineering pain points you've hit over the last year what do you see is the the kind of things that are still holding us back nothing comes to my mind oh it's all fantastic then moving on Leon you do the easy projects do you? maybe something that could help to have a real world map of different projects to things to try to establish plans and maybe predict what's going on but as I said before we adopted open source way to do things so this conference is perfect to know what is the state of each project at a given time fair enough, thank you I'm not sure it's answering the question it's good enough, thank you engineering approximation to an answer Leon in my opinion I have the feeling that sometimes we're losing too much time in politics in finding the the right words in polishing presentation instead of writing software speaking about SVN I remember that when I started working 10 years ago I was using SVN I was making core telephony solutions and mobile applications for Symbian and you know what happened with Symbian and the company behind it because exactly 10 years ago the game was changed a new product evolved on the market and nowadays I try to think about vehicles about cars, electric bicycles and that kind of stuff like a computer with wheels and I believe that this is the game changer nowadays it's a people still think about the vehicles but actually they're computers and another negative thing that it's more engineering is the problem that both AGL and Geneva Development platform are having with the licensing model in Qt because recently Qt changed the licensing model to GPL and LGPL version 3 starting as version Qt 5.7 and both platforms for the moment are using the octal layer with Qt 5.6 I hope that this is something that is going to be solved in your near future and we'll find a way to move forward Okay, thank you Leon So you're echoing what Agustin said that the license actually is going to become a bottleneck in a way, okay Arthur? I think our biggest pain point and this isn't a huge surprise is just the scale and the amount of legacy in these organisations we're trying to push solutions to problems and we need to make them work on platforms that are five years old or on microcontrollers where they can't spend an extra penny on a bit of extra flash and so we have to have solutions that scale all the way from the most modern Linux systems to an 8-bit microcontroller that's been in a system for 20 years and that makes it very difficult to deploy any kind of modernisation Yeah Okay, but Stefan? I would say that, yeah I agree with that, yeah The scale is quite large so covering everything is maybe difficult There is also some things the continuous integration etc is a time eater I don't know if it's the world we eat our time with the Yachto migrations and stuff like that but I think that's life I would say that I would like to see less politics more code and more people involved in the code more bugs Definitely because that would mean that we have requirements and that we try to follow and we try to solve bugs to follow those requirements so we lack of sometimes of requirements in maybe some areas or domains where there would be some differences between competitors I mean for example for the other systems or systems in the automotive we don't have many requirements so we are a little bit blind and we don't know where to go and we know that for example we need some hypervisor as I said before but what will run on the hypervisor will we use KVM or Xen or whatever what would be chosen by the OEMs we don't have a clear vision I think that is a lot down to the history of automotive where obviously there's a tremendous body of engineering that's been done but most of it's still proprietary and there are lots of people who cannot participate in these conversations because of their NDAs or just because of the culture that they're in The paradox is that the OEMs ask us to create systems that we could have great for 20 years or at least 10 years but they don't even ever know what they will do in 20 years so how can we create a system or think about something 20 years before it's not that easy again can I throw it to the audience anyone got any comments about engineering pain points that they see in automotive you're much shyer and more quiet than last year there was a guy Luke Fahargan who stood at the back of the room and some of you know Luke obviously are laughing and we'd been talking for the best part of 40 minutes and he said you guys aren't living in the real world he said you're doing all this stuff in the open and that's fine and he's an open source guy big reputation in open source but the fact is on my production project none of what you're doing is relevant at all please go ahead it's coming back to my point before what you were mentioning so you always talk about a little bit bottom up so we need a hypervisor we need these upgrades and the problem is that these car vendors they want to build something real bigger that's more than the device in the car there's a cloud, there's a fog maybe and again and asking for requirements so in the best case even requirements across car vendors that is asking for standardisation yes but can we get those requirements into the open? yeah but that's what I'm coming from the telecommunications industry and we have seen in the long time I have been working there replacement from hardware to software so we have already hundreds of millions of software lines of code of software and this transition could scale only to this amount by having standardisation on the function level not every the detail so the car vendors have to differentiate of course but you need to on the high level architecture level you need the standardisation otherwise and that's something time is ripe now is exactly on your way of life I think he agrees with you completely actually I'm resuming what I was saying just before and you asked me the question so it's official now you know about maybe the open interconnect consortium open interconnect so this is a group of industry we want to standardise different domain and they want to prevent a silo in IoT some let's say there is no reason than the the technology in your washing machine is different than the one in your car when you can have a common language and something standardised so for your information I think one month ago a group has been created called the Automotive Working Group in this and our mission is to to define a common model of the car and then how to to interact with it inside the car with home devices or even at the city level for instance or even you can think bigger with a cloud and all kind of things but as an engineer I prefer to think smaller and to see if it can scale and you can imagine something bigger if you start thinking about so many services and you need the software to be set up to start doing something usually if you're lucky it happens if not you're switching from one project to another but at the moment we already made some demonstration and I can share some resources but I think this can sound stupid but the definition of what is a car is something interesting for standardisation and especially if it's not done by one company, several companies because people who have the money are the ones who are deciding so if they put their money in a common group then we can expect it to make it real OK so I've got broadly two things I want to cover in the remainder we have about 12 minutes left I think each of you gets a couple of minutes to spell this out so I'm going to say two things I'm going to also open this to the audience last year we had a few key messages which I undertook to take back to members of the automotive industry and members of the community it didn't go down well when I took them back but I will summarise what I tried to take back so on last year's discussion we identified that automotive should stop thinking it's special because it's not any different from some other industries that have equally difficult challenging requirements telecoms will be one, aerospace, defence trains so there was a general feeling from the engineering community including the panel that we should try to avoid reinventing the wheel in automotive and there was as Luke highlighted but then other people commented a fear that real production projects weren't making very much use of the public work which called into question how relevant it was and whether there was enough connection and in a way this speaks to the comment earlier that product and platform are different and maybe we need to bridge that gap so my two questions to you and you can choose either one has anything changed or is there a different message that needs to be taken back to the community this year or do you agree with that same one and if you had the chance to call to action what would you want to change what would you try to get changed so first of all to the audience is that the same this year or have things changed are there lots of people with production projects based on AGL and Geneva that want to stand up and shout or do we still think automotive is too special for its own so that's pretty much my opinion automotive is very special because these things are highly mobile very fast can be very dense and many of them out of them of course for other sensors we will even have two magnitudes more out of them but there are many cars and just looking on the use cases for connected cars they get really tricky we need to research a lot invent a lot to get this done to get the feedback loops for the control across several cars to get that done I take your point and I'm sorry I'm going to cut you off just because I want hopefully to get one new person to speak yes gentlemen at the back please and thank you for helping running that microphone you're doing a fantastic job so my point is the talk was about specifications and creating a platform and I think what really would drive this would be an OEM to stand up and say okay I'm going to do a real product with for example AGL and here are the requirements here is the time schedule and I want to go SOP in I don't know five years or four years does anyone care to say that that's already happened Walt? nope and I know that that's not quite how Geneva came about but that's a very good point okay so in strange order Leon either a call to action or some change in the message from this year well regarding call to action I totally agree with the gentlemen in the back OEM stand up specify what you want and make products the other cool thing from community point of view that I would add is that it would be great if there is some kind of opportunity that AGL and Geneva development platform could become more useful for individuals like to have an HMI that just works with basic use cases and people can grab it build it and put it in a kind of their own vehicle on their own risk because I've heard some people willing to do it yep okay great thank you Arthur well my call to action I guess it's a bit about not reinventing the wheel again but really automotive companies need to stop competing to develop obvious things that other people in the market already have you know we talked to people especially on the ITA topic who are super proud of their system which makes them almost as good as Tesla so they're going to keep it a secret and they need to stop doing that they need to admit that they've lost that battle work with their competitors so it's a system that makes Tesla's advantage irrelevant and then move on to differentiating their product yep that's fair okay Stefan the key message for me would be to for the OEMs and TO1s would be to participate more and give use case requirements anything that would help to envision the future and I would like to say to the community also this is a unique option or way to participate to some open source projects I don't agree with you automotive is not special at all there are some specificities people die in higher frequency if we get it wrong yeah but there are other ways to die so as well I mean so of course automotive has any domain there are specificities but if you recall what I said earlier I said that we want to create the base and a Linux kernel is not specific to automotive in fact we won't put much inside inside it which could be specific to automotive so basically if we have a community of students or researchers, other people even as work and embodied world and for example we live in Brittany and there are a lot of selling boats and marine business and we are in contact with people which are interested but what if we do an HGL to just grab the BSP and the base system to make it work on boats so that's the you have the same security problems we'll have the same connectivity roughly maybe the security the safety on the sea is maybe better than on the roads currently navigation is sometimes easier sometimes more tricky especially in Brittany so okay you have commands and differences so but be sure that if someone participates to for example OS3 project to make a better upgrading system the whole embodied world would benefit from that not on the automotive or aerospace or whatever before I come to you last word can I just do a quick show of hands who does think automotive is special you do right it's about 50-50 who definitely thinks it's not special from the point of view of the software we're doing so it is literally split 50-50 this time that's very different from last year because there was quite some fierce pushback from the audience I would just say that the main thing that I've seen a lot of press about and talk about over this last year is autonomous software or autonomous driving and I do think that makes cars different from lots of other use cases now because the main thing I fear in terms of the difference between automotive and other industries is we seem to be racing towards some very high risk technology without getting all of the fundamental straight and I take your point looking at telecoms it is a relatively mature thing and normally the worst that can happen is that a phone call gets leaked and people get free calls it doesn't by default if we get the technology wrong start killing people so I think it's different in that respect just in the scale of risk that we are bringing upon ourselves by going higher and higher into this fancy functionality Philipp I'm going to give you the floor to finish so you have two things to talk about either things that have changed since last year or a new message that needs to be taken back to the industry I think the key message I'm taking from the discussion so far is we would like to see more participation from OEMs so just a quick show of hands how many people in the audience are from OEMs one sorry hands up again sorry two two and a half okay so it is but there are some okay so we want to see more OEM participation I personally still want to see get so drop your SVN rubbish and I would say my key pain point over this last year has been clear case Philipp call to action okay so if the automotive is special if not maybe but it is special I can testify that the software is not that special and the good thing and I think it has been improved over years that those projects are really open you can grab the code just read the documentation online on wiki get online support on mailing list RIC and so on and you can build your stuff like you will build your Raspberry Pi operating system for your OEM system whatever we are not that special we are using the and to my experience those people are open to people and they can provide the feedback in the past maybe it was different because development was on behind door and this was a code release at the right time at the given place and so on now it's more the case every code is pushed online on reviewing system in all time you can discuss on mailing list so that's nice if I have a message to say that you have to now to prove it's not to do more of that yeah okay great thank you does anyone from the audience care to close the session by saying something new and different and exciting yes sir hello hi I am I work for Collabora and we are a consultancy company and we have also an automotive distribution we know who you are go on it's called Apertis and I'd like to invite you guys and people in the audience to check it out Apertis.org and maybe we can find some common ground that we can share or improve or processes development software whatever okay so just a simple question for you coming back to the not reinventing the wheel the idea of reinventing wheels is everybody else so don't take this personally but why did Apertis happen when there's both AGL and Geneva I think there was not any base system usable at the time for coming from Geneva I'm not sure about AGL the history about that I know there was some at the time was Migo and then Migo closed the business because the manufacturer behind it shut down and then there was some something going on and 20 seconds, come on there was this manufacturer needing a base system so they just decided to go ahead and implement something because they need something thank you for the answer and thank you for mentioning Apertis you haven't got any time but if we can throw the microphone at you I would like to say one more thing is it automotive is special because it's a safety critical system true, yes, agreed Walt, come on a year ago we had a GDP at the time demo platform that was completely languishing it was going nowhere a year ago AGL had virtually no commits to its get repositories we had just started so if you look at the progress that we've made AGL from a year ago it's just tremendous in my opinion and that really is the biggest change from a year ago we have a lot more collaboration going on industry-wide and that would be my final point Walt, I appreciate it, thank you very much for saying that folks, it's been great, thank you one more Pavel hasn't spoken before, let's get Pavel speaking I owe you a beer, young man thank you very much to nobody on the panel we will take some please, don't but I believe that we can talk about success of open source and automotive when those people on the panel are working for OEMs, tier 1s and at least 50% of the auditory is also until then we are just trying to push but we are still not there yet to be fair I didn't try at all to get OEM or tier 1 members to the panel it's just easier speaking to people who are very active and visible in the community so literally all of these guys have been on the mailing list on IRC but I completely take your point and if this runs again next year I think with the help of people like Walt and some of the other people I know it should be possible to get that the panel could have very much more an engineering representative from several OEMs and tier 1s but I think actually that's a very good message that I'm willing to take back if people are going to attend for next year and on that note we really are out of time I thank you very much for listening and participating and hope to see you next year, thank you