 Please welcome former DPL and current Chief Technologist for Open Source and Linux at HP with Agarvi who will be talking about the relationship between HP and Davian and some thoughts that might interest the project. Please remember not to walk in front of the camera, ask for a mic for questions and look at the camera for speaking. Thank you. Thank you very much. So good morning. I'm impressed that this many of you got out of bed to come and see what I have to say. I'm sure other fellow, they're rushing in now to see what's going on, but I do appreciate it. This talk is a little interesting because it's not something that I originally proposed doing here. There was instead a decision made to invite some of us who represent some of the key sponsors of DevCon for who are involved in particularly interesting aspects of the project to give some of these opening session talks in the morning and Marga and the rest of the team invited me to do one and I said, sure, okay, I'll be happy to. So what I did last night was to sit down and pull together some material that tries to cover a little bit of the history of the relationship between HP and the Davian project that talks a little bit about what HP does to try and support Davian today and how that looks to some of our customers and what might be interesting to you about that. And then I'm going to end up by presenting a few thoughts. I have, I guess, three different things that I thought about that I think are opportunities for Davian to do more interesting work that perhaps many of you haven't thought about or don't recognize the potential value around doing some simple things. And so hopefully by the time we get to the end and it's time to take some questions, I will not only have regaled you with a little bit of history but have also given you some interesting ideas about things we can work on together for the future. So to start with, how many of you have not heard me speak before somewhere? Yeah, right. Okay, well, I learned a long time ago that it's presumptuous to walk into a room and assume that everybody knows who you are and why you're there. And I sometimes stun some of my fellows who work in the corporate world by spending as much time as I do trying to explain who I am and what I do and why that might cause me to be somebody interesting to listen to. Since so many of you have heard me before and I think know me one way or another through my Debian history, I won't spend a lot of time today on my own background except to point out that I do recall making my first personal contribution of source code to a free software project in 1979. This is a little bit of source code assembly language for a then obscure and now completely forgotten microprocessor that got published in the newsletter of a local users group for that processor in Canada. And in fact, I was invited by the president of the club to go and be a speaker at one of their monthly user group meetings. Unfortunately, I wasn't old enough to have a driver's license yet and my parents didn't think that going to Canada to speak at a local users group meeting was all of that reasonable a thing to do. But it gave me a very early understanding of the fact that if you did interesting work and were willing to share that work with other people that really interesting things might happen as a result. I got involved with the Debian project in 1994 and I'll talk a little bit more about that in just a minute. I've stayed involved with the Debian project continuously since 1994 though there have been some times where I was distracted working on amateur satellites or having children or other little things like that. Today I do serve as some of you heard last night in our SPI BOF, I serve as president of Software in the Public Interest which is the nonprofit organization that provides financial and legal services to Debian in the United States. At the moment I'm serving as HP's representative to the Board of the Linux Foundation and in fact that's why I couldn't get here until Saturday. There was a Linux Foundation board meeting in San Francisco on Thursday that I had to go attend on HP's behalf and so instead of being here at the beginning of DEB camp I got here at the end which was a little bit frustrating. And the other thing that I'll point to that is fairly interesting particularly for this audience is I recently agreed to join the board of another nonprofit called Open Media Now. This is the nonprofit that has been started by Rob Savoy and some others to provide some infrastructure around the work that he's doing on the ganache free flash replacement and in general to try and push towards a world where we have better access to open standards and open source implementations of codecs and media tools and so forth. Those of you who have known me for a while know that this is another way to think about me. I've worked on pieces of amateur radio satellites that are in orbit around the Earth today but more recently my son and I have gotten very involved in building and flying high powered model rockets. That particular rocket is about a three inch or 75 millimeter diameter airframe as you can see not quite three meters tall. And in the first weekend of May this year in the northeast corner of Colorado in an area called the Pawnee National Grasslands which as you can see is very much like this part of Argentina big and flat and grassy I successfully launched that rocket to an altitude of 14,141 feet above ground. That's a big deal. That's a big rocket and it's lots of fun. One of you that can do math at this time of day better than I can that's 4,000 plus meters. So I do the math, I don't know. That's the motor that was in that rocket filled with an ammonium perchlorate based composite propellant that's very energetic. 4,002.3 meters. Oh dude. I love people that can use technology at this hour in the morning. So anyway this is one of the things that I do for fun and now I don't know how well it shows up on the projector but this is the new airframe that I'm building I hope to fly in September to an even higher altitude and to qualify for the highest level of high powered rocketry certification in the US. Those fins are made out of a carbon fiber composite and if you're interested I can spend hours explaining to you how to do carbon fiber vacuum composite construction using a kitchen food saver appliance and other tools that you find in the common American household. So this is the kind of stuff that I do for fun and in fact a couple of weeks ago at the Ottawa Linux Symposium I gave a full talk on the open source avionics that I've been working on to fly in some of these rockets and if there's sufficient interest here you could twist my arm to pull that slide deck out after dinner some evening perhaps and talk about it. I'm not hard to convince to talk about these things. But I'm here for something else this morning I have to keep reminding myself. So if you don't remember anything else from what I have to say this morning this is what I would like you to remember and that's why I made it the topic of the talk. hp.com.go.debian is the portal homepage whatever you want to call it for finding information about the support that Debian receives on products that are produced by HP. Typical of corporate websites every once in a while there's something new that's happened that the website hasn't been updated for but this is the thing to remember if you want to go find out where to get bits and pieces of device drivers and so forth to use on HP products that are running Debian. But how did we get to there? Let me step back just a little bit and talk about the history of the relationship. Some of you may have heard me talk about this at other deb-cons in the past but it's been actually at least two years since I talked about this in public so I thought maybe it was a good time to refresh this a little bit. Back in the early days of the project, around the end of 1994 I noticed Debian because I was very involved in an amateur radio satellite project and I needed a development platform and without boring you with all of the highly amusing little details at the end of the day I discovered that my friend Bruce Parents was working on a project called Linux for Hams meaning ham radio operators and in order to do this he had discovered and become involved with the Debian project and was helping to create an installation tool which became boot floppies so that it would be straightforward for people to install this new Debian distribution which he then hoped to use as a basis for creating what we would now think of as a CDD a custom Debian distribution aimed at ham radio operators and so I thought oh that's cool I'll help out with that and I got involved a little bit in the fall of 1994 installing an early version of Debian and I think it took all of about 48 hours for me to notice a couple of utilities that I thought were important that were missing and to offer to provide those to the project I've chuckled at various times in public that my new maintainer process was Bruce Parents sending Ian Murdoch a message saying this guy BDL that I know from the ham radio world wants to help and Ian sent back a message welcoming me to the project it was a very different time but around this time Bruce was very concerned because the main server used by Debian was this thing called ftp.debian.org and that was actually an alias on a PC that was under some students desk at a university somewhere in the middle of the US the details get a little vague at this point but Bruce was concerned because every time anything happened in Debian the university's one or two T1 links to the rest of the world would get saturated with people trying to pull updates off of this kid's PC under his desk and he was afraid that at some point the university might just shut it down and you know what would happen to all of this goodness and Bruce knew that at that time I was working in a job in the old test and measurement part of HP where I had a team of people maintaining a whole pile of servers and we had lots of resources and he asked me if there was something I could do to maybe grab a copy of all the bits in case something bad happened and to make a what could easily become a very long story a little bit shorter I scrounged up a cast off HP Vectra 486 33 megahertz tower system that was something that was being thrown away that gives you a general sense of where we were in the technology history and I put two SCSI drives on it a 300 meg disk for a root system and a 660 meg drive that held the entire archive plus lots of space for growth and we put Debian on it I think that would have been 0.93R4 I'm not certain but I do have the stack of floppy still in my basement at home I am a pack rat about things like this who knows maybe someday the Smithsonian will want to have them or the computer history museum I don't know and that was the beginning of master.debian.org which was the first time we had a separate dedicated machine for building the Debian archive and then we used ftp.debian.org and an ever growing set of machines to replicate that and make it available out in the world I will point out that as a manager at HP at that time responsible for that set of resources this was a decision that I was allowed to make but it was to my great amusement some number of years later when HP marketing discovered this little bit of history and started talking about how great HP support had been in 1994 so if we fast forward a little bit in the middle of the 1990s the PA risk architecture from HP which was arguably one of the higher performance risk processor families in that era was the only major 32 bit CPU family that wasn't running Linux while there had been all sorts of interesting discussions in HP about what to do about this a little group of consultants in Canada calling themselves the puffin group made a big name for themselves by announcing that they were going to begin a port of Linux to the HP PA risk processor and I don't know how many of you remember but back then there was a conference series called the Atlanta Linux Symposium or ALS and one year ALS sponsored some incubation rooms where people with good project ideas could come and have some floor space and some bandwidth to get together and work on interesting projects and the puffins got access to one of those rooms and over the course of ALS successfully got to the point where with an amazing amount of completely ugly hackery it successfully booted a Linux kernel on a PA risk processor to the point where it had a shell prompt and at that point somebody at HP said oh gee, they made a lot of progress really quickly and maybe we should help and so HP agreed to participate this was an incredibly big deal for those of us who worked inside the company because up until that point I think there had been this belief that we needed to be really careful about this Linux thing because after all the company we worked for produced a commercial Unix operating system and by the terms of our employment agreements working on competing technologies was considered a conflict of interest but all of a sudden if HP was endorsing this port of the Linux kernel to the PA risk processor and had agreed to make documentation available and even to allow some HP employees to work on it all of a sudden Linux was okay to work on as a company employee and people came out of the woodwork it was one of the most amazing things to watch from inside the company because it was an empowering event that all of a sudden these employees felt that it was okay they still understood they had to talk to their managers they couldn't spend work time doing things that weren't related to their job and so forth but it really represented a switch flip in the company where people decided hey this Linux and other related open source things must be okay for us to work on because the company is agreeing to support it and then business happens Linux care acquired the Puffins this was no longer just a fun project the Puffins were working on to make a name for themselves HP was approached about a contract to pay for some of the work that was being done and at some point HP woke up and realized that having a Linux kernel on a tool chain was kind of cool but it wasn't very useful unless it was part of a complete distribution the problem of course is that the public message had already gone out that PA risk as an architecture had a finite lifetime that HP was joining with Intel and with some other companies to put a lot of emphasis around the itanium or IA64 architecture for the future and so commercial distribution vendors looked at the PA risk work and said well that's kind of technically interesting but there's no money in it because you're not going to be making these machines forever there's a lot of money building a distribution how many of them would we really be able to sell and would we ever make any money and so the commercial distributions wanted really big dollars from HP to go do the work to create a distribution and because this was largely a community activity and not something that was being directly driven by HP having a business need for a distribution it didn't make sense to anybody to spend big dollars producing a Linux distribution however a bunch of engineers who were in and around this work in HP happened to have discovered Debian some through my cajoling some through other paths and we sort of floated the idea that well hey you know we could just enable Debian to be successful at completing a port to this architecture and then we wouldn't have to spend steaming huge piles of money to get it done and that's what ended up happening there were some very conscious decisions made in those early days about how the company would engage with the Debian project the most important was that this wasn't going to be a fork this wasn't going to be a situation where we took a copy of the code and went and built something and gave it out to customers that instead we wanted this to be a real Debian port we would engage directly with the Debian project and the various upstream maintainers to make sure that all of the work that we were doing went and was merged upstream as quickly as possible we hired some existing Debian developers to work on the project we encouraged some serious HP engineers with long-time experience in HPX to come over and join this project some of whom became Debian developers and are still to this day and we made some selected investments in helping to further certain Debian features one of the most interesting, very small but I think important investments that we made in about that time was the agreement to provide funding for the legal opinions that allowed for crypto software to be moved into US Maine if I remember correctly that was just prior to the Sarge release and it was a really big deal because all of a sudden Debian didn't have to sort of have the crippled version that we could find lots of places to distribute from in the US and the full version that you'd only be able to get in the free world and I don't know, I was very proud of the fact that we talked HP into spending that money by the way it was a tiny little bit of money I think it was less than $2,000 US it was not a significant amount of money but it was another significant commitment in the willingness to help the project get over a particular speed bump and all of this experience that HP's engineering teams gained from working on the PA Risk Port in how to work with the Debian project and how to work productively with other communities the Kernel community, the X community we actually have been very involved in X for many, many years as one of the original companies involved in sponsoring that work but all of these different open source communities this PA Risk Port actually was a key enabler for us to learn how to work with the community and how to behave in a productive way and it's led over time to lots of other work being done using Debian, taking advantage of Debian contributing to Debian in different ways and not surprisingly along the way we've convinced ourselves to make a few hardware donations to help the project out we'll leave it to others to rattle off how much, but it's a lot there's some other really interesting things that have happened along the way a few years ago HP got very involved in producing a flavor of client called thin clients how many of you have played with stuff like the LTSP software the Linux terminal server project these are not really like LTSP clients they can be used that way for those of you who haven't used it LTSP is software that allows you to take any client system with no local operating system image but the ability to boot across the network and boot it up running a full client Linux image that can make that hardware very useful and more manageable than if you had to have separate operating system images on all the machines and so forth these thin clients are similar but different the difference is that they really do have a significant amount of flash memory that looks like an ATA drive in the box however they have no moving parts there's no fans, no hard drives, none of that stuff so they are super reliable pretty low power, they're not as low power as the board that Mad Dog's walking around with or some of the other more recent systems we've seen but these have been incredibly popular in interesting places like stock exchange trading floors places where the reliability of a client system is really important the fact that this little thing can be screwed on to the back of a LCD panel in the whatever it's called the visa mounting, screw locations and so forth and the highest powered versions of these boxes and the lowest power, most integrated most black box-ish versions all run Debian and in fact that team is very enthusiastic about what using Debian has allowed them to do it's allowed them to enable the end users to customize the configuration of these systems very easily to add applications to remove them I clipped the chunk of text in the second bullet here in the wee hours of the morning this morning directly from the HP website where these are sold and you notice it's talking about this powerful Debian Linux based client and features include Ice Weasel and you know I get a certain pleasure in seeing things like this showing up in marketing content on websites but these thin clients have actually been hugely successful and even today it's a non-trivial I mean it's still a relatively small slice of the total client market but HP's position in that marketplace is very strong and I don't think it's despite using Debian I think it's because of some of what's been enabled there if we look at the title of this talk hp.com-desk-go-slash-debian this is a consequence of an announcement that we made in August of 2006 at Linux World in San Francisco where we initially announced that we would support Debian on HP's ProLiant server family even today to the best of my knowledge this is completely unique within the industry of first tier hardware OEMs no other company provides the same level of explicit support for Debian running on their servers and this has had a very pleasant consequence for us from a business standpoint particularly in European countries where we've seen much more interest in Debian than in some other parts of the world if you go to that web page and I apologize to the folks looking at this later this is going to be sort of lousy looking on the video stream I'm sure this is another thing I grabbed from a web page last night if you go to hp.com-go-slash-debian this is what you see nice little text about the Debian project being a well respected strong project in the open source development community and down in the bottom some links that will take you off to the servers, clients and the services that are available it's sort of interesting I think that this work has been very successful in helping to motivate people to feel good about buying HP servers to run Debian but there's been very little response to the offering of higher level services for Debian and this is something that I could actually use all of your help in thinking about through the rest of the week in various conversations that we might have one of the things that we have to make really smart decisions about at HP is how much work do we really do on behalf of something like supporting Debian and that has to be driven on the basis of what kind of financial return do we think we're going to get and one of my observations has been that people who choose Debian over some other distribution are often the kind of people who don't want or need to have the higher levels of telephone support that might be provided by a company like HP and so forth and because some attributes of Debian make it not as desirable an operating system platform for people running big commercial applications as some of the other distributions and those people are often the ones who are willing to buy things like a seven day, 24 hour day support contract I think it's important for us to understand where the right threshold is between the kinds of services that we try to provide that will actually be useful and meaningful to customers and providing all of the options which is very expensive for the company to set itself up to be able to do and may in the end really not be helping Debian or Debian's users very much so I would encourage all of you to think about that a little bit if you are using Debian on HP servers in a commercial kind of way in support of yourself or your customers please come and talk to me about this and let's have some interesting conversation and see if we can't come up with a good thought about how much investment we really should be making to meet the needs of Debian customers without overextending the company's investment because of course the fear is if we spend too much money doing too many things that don't look very successful then that's not good for the long term relationship between HP and Debian so I want to talk about a few specifics of the current status and future support for Debian on the ProLiant servers there is a commitment to update all of the current Debian driver and agent and so forth support for ProLiant once Leni is released there is an expectation on the part of the people planning to do that work that means September sometime and we'll see what happens as always happens our schedules get adjusted it's worth pointing out that for a while to get full functionality on the ProLiant server family has required running some HP specific drivers we're working very hard to get all of the required functionality included upstream so that you don't have to do that the HP ILO driver ILO is the integrated lights out the remote management interface for these servers that driver was merged upstream for 2627 and I think partly due to Max's work that's been back ported into 2626 and hopefully will be where it already is in the builds that are in Leni I haven't had a chance to look there was some functionality that was unique to HP servers in the open IPMI interface space that required use the separate version of the open IPMI driver we're really pleased that that's all been merged upstream now so you shouldn't need the separate open IPMI driver going forward and near and dear to my heart we had this package called HP AOSM that was sort of a mega package that included lots of SNMP agents and so forth and buried down in the bottom was the one truly useful little bit to me at least of the speed control for the servers and I'm pleased to report that Dan Frazier tells me that's all been split up now and I don't know if the new versions have all been released yet or not but whether they have been or not it will be possible to have fan speed control without having a steaming pile of SNMP agents loaded and running on the system which all of these things are things we think are improvements and are responding to criticism and requests that we've had but fundamentally I'll point out that there are quite a number of improvements being made in the packaging Dan Frazier who many of you know is involved in the kernel team and has been involved in the security team and so forth his full-time job now is helping to improve the state of these drivers and some of the related packaging for support of Debian on Proliant and yes we really have in the past shipped packages that were built all of that should be substantially improved as we go forward I'm very pleased about that so to summarize all of that stuff if you don't remember any of those details just remember this you can always go there, fish around find the current drivers and so forth so now let me shift gears just a little bit I don't know what time it is just after the bottom of the hour let me spend a few minutes talking about a few thoughts that I have about the project someone who has now been involved fairly actively in the project for 14 years I guess it is coming up here very soon I have spent a lot of time thinking about Debian and more recently I've spent quite a bit of time thinking about things that are happening at HP and that might be interesting to Debian or where there might be an opportunity for us to jointly do some things the first thing I wanted to do is I get this question a lot from people what's important to HP what matters to you I always have to start off by saying first of all it's a business and it's a publicly held company that operates as a business and that means that the most important thing to HP is increasing our shareholders value generating more return on the investment that the public has made in our company purchasing and holding our stock and so you should never be confused with HP or any other large company nothing that we do is done just because it's good or just because it's the right thing to do we are fortunate that in the world today most large corporations understand that in order to increase shareholder value they have to behave in ways that are compatible with society and do things that make sense for the common good but don't ever be confused that something we do is being done because we love Debian or because we love open source or free software or something like this at the end of the day any time a large company does something it's because there is some opportunity to generate increased revenue and profit as a result of that action and the thing that matters most to doing that is the quality of the experience that we deliver to end users together and so every time someone has a ProLiant server and they choose to run Debian on it and they have some problem installing on the system because gosh, Debian decided that that firmware shouldn't be included in the kernel package anymore it's a challenge, it's a challenge for us it's a challenge for the project it's a challenge for that customer to try to load, you know the HP ASM package on an AMD64 system and have to go find and install a 32 bit library to make all the pieces work it's a challenge, that's something that collectively we need to figure out and work on I think if you notice if you've been paying attention there was a point in history where HP's behavior was pretty stoic around a lot of our products a lot of them had a very industrial and feel to them and they were known for being reliable and high performance but really not very exciting and over the last few years that's been gradually shifting you've probably seen the computer as personal again advertising campaign that our PC guys are doing and recent introductions like the TouchSmart system with the touch-based user interface for the home and there's a lot of work going on now into understanding what customers and end users would really like and to trying to provide better end user experiences we had some interesting discussion yesterday in a couple of sessions here about things that can be frustrating or challenging to people trying to install Debian or get the desktop environment that they want it's very easy for those of us who've been involved in the project for a long time who know the six magic words to type at the right point in the process to say to ourselves and to others around us oh that's no big deal you just have to do this what we should instead be doing is thinking about why was it that somebody had that problem what did they have to stumble over and what did they have to learn that maybe we could have made it easier for them to navigate or negotiate and those are all things that in the end will make Debian a more desirable distribution and will make Debian's users end HP's customers trying to use Debian more successful the other thing that I talk about a lot to people when they ask this question is that you must understand that for a company like HP there's always a balance between what we decide to do ourselves to directly hire people to work on and to make direct engineering investments in and what we expect other people to be able to do for us under contract part of the reason that Debian isn't the distribution that naturally gets included in some of our desktop notebook and many notebook products and so forth is that HP doesn't actually do the engineering work in the most part that leads to those pre-install images those are produced by other people as part of a contractual relationship with HP that includes marketing activities and support activities and so forth but sometimes the fact that Debian isn't a company and doesn't behave the same way as a commercial organization can present some challenges because the way HP has to market and support products that use Debian is just different from the way we market and support products that run commercial distributions in the enterprise server space one of the really big issues is that people want to run these big ugly proprietary applications if you don't have a really good story for how you support Oracle and SAP and other application suites like that there are a lot of customers for whom you're just not interesting just not relevant now is that something that Debian should worry about it's not clear to me in the case of Debian we are incredibly successful at being the distribution of choice in the case where people have made the intellectual leap to having much more or even all of their software be open source free software and to operate in that mode where we are not good is at trying to be just like the commercial distributions and have ISV certification and support relationships and so forth this is something that is a distribution I think we have to think about once in a while so that our decisions about how hard to pursue these things whether to work on them whether this is important to us or not are conscious decisions that we're making but we have to understand the consequences of those decisions if we as a distribution don't have Oracle supporting Debian as one of the combinations of a slice of the potential enterprise customer market that just won't be interested and that's just a fact predictability I've talked all the way back to when I was running for DPL in 2001 or 2002 somewhere back then about the importance of being predictable that doesn't necessarily mean regular and with apologies to Mark it doesn't always mean synchronized with the rest of the world though there's a lot of interesting things to talk about there what it really means is that when we make public commitments about things we have to do our best to try and follow through them because other people will build expectations on the basis of that and everyone understands that at the end of the day Debian isn't going to release until Debian believes what we have is ready to release but the harder that we work to try and make that be a predictable process the easier it is for others in companies and our users to be able to cope with us and figure out how to build their plans around it I've already mentioned briefly just persistently as a hardware manufacturer always chasing new functionality building PC clients with new chips in them building servers that have long-term commitments to particular chip sets we always have around the edges issues that we're dealing with around kernel device drivers and firmware and so forth I'm really pleased by the way with the way the Debian and Solar team has helped recently to tackle this notion of how we deal with firmware that's required by device drivers that's not part of the Linux kernel source tree anymore and I think that the solutions that are being worked on there are great but that kind of stuff is important and it is important that we have a good story for how a hardware manufacturer producing new hardware can support that and make it work with our distribution which is often going out in a very different release cycle than those hardware products another thing that HP did at the beginning of this year how many of you have heard of Phosology, Phospazar before a few of you I'm actually really pleased by that I'm disappointed the rest of you haven't but let me tell you really quickly what this is all about a few years ago HP got very serious about explicitly managing the relationships between open source software that we use in our products and that we provide contributions to in the open source community and our intellectual property which often means patents and related things of value so earlier this year we made a public announcement and have released the source code to the tools that we use internally to discover what licenses are existing in different pieces of source and binary software and to do analysis of those to look for compatibilities, legal issues and so forth this was released as a pluggable infrastructure a lot of data database with agents that know how to scan source code to discover things about that source code and then tools that know how to present the information from that database in different kinds of views into a web browser kind of interface this has actually generated a great amount of excitement on the part of people who are in the academic world analyzing different aspects of software because now instead of having to start from scratch they can just go create a little agent that looks for the things they care about whether it's complexity metrics or other legal or licensing issues or vulnerabilities unbounded use of stir copies, things like that and plug them into this infrastructure Martin Nicholmeyer, former Devian project leader was hired by HP specifically to serve as the community manager for FOSS Bazaar community of best practice among end users who are using FOSSology and other tools to manage this I think the FOSSology toolset the actual technology that was released as part of this should be very interesting to Devian and in fact I've had some conversations recently with various folks in interesting places around the project because for example the license analysis that's done here could make some of the work done in processing the new queue more straightforward and even more than that I think that hooking this into our archive update processes might allow us to more quickly and straightforwardly notice when some license has changed on some new version that's being uploaded into the archive if you think about it today those are sort of discovered after something happens we have an explicit process for using the licenses and the legality and other packaging appropriateness and so forth of things that are being put into the archive for the first time but we don't have any sort of automated process for noticing when a new upload of some existing package in the archive goes through a license change and when that happens when somebody incorporates new functionality brings in a new library we all know that sometimes we end up with unholy things that we don't have in the archive. I think this is a toolset that could be interesting if you're interested in this kind of stuff and you haven't gone please go look at fossilogy.org and have a look at this toolset and perhaps join some conversation about how we might take better advantage of this in the Debian project. The community of people around fossilogy and PhosBazaar are working to do analysis of this project. I think we want to see how we can take advantage of this to improve our own project processes not be the subject of someone else's analysis. The last thing that I want to throw out an idea about this morning is that every once in a while somebody asks me inside HP what are we supposed to do to certify our hardware with Debian? If you want to certify for the Linux Enterprise desktop there is a certification test suite that you can run. It's not all that useful in my personal opinion because for example you can pass the test suite if you have a working network interface and on notebooks that's usually passed by plugging in a wire to the Cat5 connector and the back of the machine and gee those tend to pretty much always work. It doesn't tell you anything for example whether the wireless works or whether there's 3D acceleration available with an open source driver for the graphics chip that's in your machine. I watched with great interest a while back when some folks at Intel and elsewhere produced this thing called the Linux Ready Firmware Developer Kit. How many of you have heard about this? Have you actually run it? It's actually kind of cool. It's a bootable ISO image that boots up, brings up a Linux kernel, runs software and tells you whether the firmware in that machine is ready to run Linux or not and it performs a lot of tests to try and figure that out. At HP, when that kit became available, all of the people working on products started using it. Why? Because it's easy. They can take some intern or new employee and hand them a CD and point them at a row of machines and say go try this and report on the results. It's really interesting if Devian were to spearhead the creation of something a little more than just are we ready for the Linux kernel in our BIOS firmware, but what if we would actually build a real client or server certification test that would at the end of the day only give you a big green check if all of the interesting functionality that anybody cares about was successfully enabled using open source device drivers. That would be a really interesting thing for Devian to do. It would be a really interesting statement and we don't have any conflicting expectations. When a commercial distribution produces a certification test, they have a conflicting expectation because on one hand they would like as many systems to pass as possible so that their software looks like it runs on lots of different stuff. And on the other hand they'd like to do the right thing to make sure that the things people care about work. There's a tension there, a bit of a conflict. And at the end of the day I think this is another example of where the Devian project with our lack of need to be torn in different directions like that might be almost uniquely able to produce an interesting and really useful to end user test kit like this. Do I have time to do this? No. I have my own rule from yesterday and I'm going to whine about something without volunteering to go do the work. But if some of you in the audience think this is the sort of thing that sounds interesting, I'd love to talk to you about it and I'll try to help if making some hardware available here and there would help things like that. There's things that we can do to try and work together on this. I think it would be a really interesting thing to do and I think it could meaningfully influence the behavior of the entire industry. And then it's left. I'm happy to take a question or two now if somebody has something that's a burning passion. Otherwise I'll be here the rest of the week and happy to talk to folks in the hall. Thanks for your time and your attention. Do we have time for a question or two? Okay, so questions. Phil, somebody got a mic. We'll take that one. Phil down here in the front has one too. Go ahead. I would just like to make a small observation about what you said. It would be really, really interesting to increase our user base. I work for the government of Brazil and I can tell you if we have some sort of agnostic kit we can use to say here if you pass this we can buy from you. If you cannot pass this we are not going to buy from you and that thing does not smell like it was done by a special company to get their products. We can use that for government purchase and government doesn't buy small and if we start buying too much of IBM that means people start making machines and software that works on IBM and that doesn't work just for Brazil, it works for all governments. That's something to keep in mind. It really is our best interest to have something like that and also it will help to get yourself a better understanding that works properly. I will admit one of my great pleasures is that I now have a notebook on which everything works right in Debian and it's very cool. To get to this point even when I have the job that I do with the company it's been tough. Anything that we can do to make that better would be a good thing. I was puzzled and intrigued by Oracle's decision to do the white boxing red hat thing because they are having to do some support for that and they still have to deal with the fact that red hat occasionally do things which break their software on top of it because they have at the last second inserted a patch to the kernel or whatever. Why is there any way that we could encourage them to have Oracle Linux based on Debian because there would be a lot less work for them and they wouldn't get stitched like that on a regular basis. The problem is I don't know that it would really be a lot less work for them because the reality is they have already done all of the work to enable and package their software on top of RHEL and by keeping OEL in fact RHEL 5.1 at least I did a very detailed analysis and know exactly what they changed in every piece of source package between RHEL 5.1 and OEL 5.1. The changes are much smaller than they were initially so the behavior I see is that they are actually trying to sort of minimize those differences so that they get the most leverage between their existing commercial supportive RHEL and the work they have to do to support their own distro as possible but they don't currently to the best of my knowledge provide direct support for any Debian drive distribution I'm afraid it would look like a new port to them so this would be a wonderful conversation to have offline I have lots of thoughts about the pros and cons of the existence of OEL and the ecosystem and at the end of the day I don't know that it's the move that I would have made at the time they did but given the existence of Oracle Enterprise Linux and the impact it's having in the marketplace it's certainly something we ought to think about and understand so I gather we're pretty much out of time now I really appreciate all of you getting up this early to come and listen to me if you have other things you'd like to talk about come and find me and we'll have conversations and all was there somebody else that had a question yeah yeah go ahead all questions are short it's the answers that are tough I should say all good questions are short yes what's the opinion about the HPPA and ALFA port HPPA port is mostly dying because many of lack of manpower and I guess the ALFA port will the same will happen for the ALFA port I don't think that Debian should apply any special criteria to those architectures because they happen to be of interest to HP I would like Debian to apply the same criteria to the ALFA and PA risk architectures in terms of deciding when they have run their useful life that they would apply to any other architecture and I won't try to announce right now what I think those answers should be but I'll be happy to have conversations about that if you'd like but what's important to me is that that rule that we began with this is not some special port this is just a real Debian port should continue to apply until the end of the life cycle we should not extend it longer just because it's HP we should not cut it off sooner just because it's HP let's treat those architectures the same way we would treat any others in the distribution okay, thank you very much