 volunteers approached us which would like to make a nice system containing, so open the large kernel and some parts of the Debian user land. I don't know how many males were exchanged in that flame war. I would say both sides make some small mistakes. So it's a nice, it's a pleasure to have now here Simon Phipps and several others who'd like to talk a bit about it in a civilized way. I'm sure and see if we can be friends. So please welcome Simon Phipps, Don Alvaro and Mark. So hello everybody, stand up. So I'm not actually from the group that joined in with the polite and friendly discussion that was on Debian Legal about NextCenterOS, but I did watch with great interest when it took place. So I work for Sun and my job is to, I'm responsible for Sun's involvement in free software. And yeah, thank you very much. Also Sun is under new management at the moment, I don't know if you spotted that. And so my personal opinion and I have to be very careful how I express it particularly with the camera rolling is that Sun's management made a few small mistakes over the last few years and said some things that I don't think were very well thought out about free software in particular. And for a number of years I was working with Denise Cooper who's at the back over there. And now I'm working alone because Denise has run off to an even bigger and more evil corporation than the one I work for. And so now I'm left with a new boss trying to work out how can we fix the problems that Sun has had with the free software community. And I always think the number one thing you have to do when you do that is to show up and start talking to people. So I'm here and I want to talk to people. Second thing I think you have to do is I think you have to find tokens that demonstrate that there's no hostility. So I went and stole a T2000 server from our NSG group last week and I gave it to Debian last week. So that's installed up at Oregon State University. And if you want to do builds on Spark it's sitting there waiting for you now. If you talk to Martin Mittelmeier he'll tell you how to get onto it because he's looking after it for us. So I want to do things like that that are partly symbolic that are indicating a lack of hostility. Now then the next question is what do we have to do in practical terms to work together. The way that I was talking about this with a colleague last week was that I feel like I'm wandering around a neighborhood trying to arrange the neighborhood watch. And we've got all these street gangs coming in from the Pacific Northwest of America who are fouling everything up for everybody. And so I'm going from house to house now saying look let's have a neighborhood watch. And I don't want to comment on things that don't lead to good relationships. So if you're doing the neighborhood watch you wouldn't point out the fact that it seems to be the perfect household that people live in but no one talks to each other or the color of people's hair or whatever it is. You talk about the issue at hand. So now the issues at hand as far as I can see there are two big issues that really concern me in working together with the free software community. And one of them surrounds Unix and Linux kernels and the second one surrounds the Java platform. And I want to fix both of them. And let me tell you what it's like working for a corporation with 38,500 employees. In a corporation that size no one calls the shots. It sounds really strange but let's say that I persuade Jonathan Schwartz tomorrow that he wants to take the Java platform and release it under a free software license. Well he can't actually make it happen because he's delegated that responsibility to someone who works for him and that person's delegated the authority to someone who works for him and they've set up all sorts of financial relationships that they have to tell to financial analysts about what's going to happen in the company. And if suddenly on a coin it changes strategy and it's all different, well it just fouls up the whole company. So I can't just snap my fingers and make it happen but I am desperately keen to make both of those situations get fixed. So I'm coming here and the context in which I want to talk is what is it that I've got to do in my job. I'm getting you along as my focus group. Okay. What have I got to do in my job to turn Sun into a company that makes free software succeed? What have I got to do to make that happen? That's what I really want to find out. How can we be friends? And if I leave here from Debcom knowing what it is I've got to do to make Sun into a friend of free software then I will be satisfied. So that's the context that I'm in. So now I've already got a question on the back over there. So we're going to try and wrangle these microphones. Thank you. This one does. Yeah, well in Denise's talk yesterday we heard that basically the big difference between Intel and Sun is that Intel is mainly selling hardware and Sun is also trying to sell software. Is this going to change or? So no, that's not going to change. What is going to change is the way that Sun thinks about selling software. Are we selling services? Well there's services and there's services you see. So if you've got corporate customers what it means to sell them services is to say to them if you find something's wrong in the software that you're running your business on I will fix it for you and I'll guarantee you that the fix will end up in the next ship version of the software. Now that's the service that corporate customers actually want and Sun intends to sell that service. Yes, Sun doesn't intend to sell consulting services or education services or configuration services or call center services or all that sort of stuff. But we absolutely intend to sell bug fixing services that involve paying committers to work on software. So one of the people I've got on the group up here is Alvaro. Is that working now? So Alvaro works on GNOME. On the JDS team. We'll just mention about that. Don't mention JDS, don't mention JDS. Well I'm working in GNOME and Open Solaris making it work together much better than what it do now. And you know right now we are working in the community we don't do anything inside Sun. Everything is you know using back sea line in GNOME for example. And it's really amazing to work in that way. I mean we've never done anything that has been hidden inside Sun related to GNOME or anything else. So sometimes when people you know tell me about the behavior of Sun inside the free server community it's kind of you know weird because we are doing everything outside. It's like that. So yeah we'll sell services but there's a particular kind of services. The basic question was about whether there's a business model that supports that. Since I'm also a happy Sun customer because I spent a lot of money on hardware and it got Solaris for free. So you're obviously not selling Solaris as a separate product. And you're involved with free software. It sort of requires well you can basically sell free software but people the first thing that's going to happen is that people fork it. So probably and well as I said in Denise's talk we heard that basically that was a big difference between Intel and Sun. And at the moment I don't see the difference. Yes. So at the risk of having a discussion about Intel and Sun you know the big difference is that you actually know that we sell software whereas you've no idea how many software engineers are working for them. And you'd be shocked if you knew. But ultimately I think that the future of computer systems selling involves giving customers packages that work and in the world that is increasingly dominated by free software giving people packages that work involves employing people to make sure you can promise to fix bugs. And those people get the ability to promise to fix bugs by being good citizens in a developer community not by being appointed there by a company. So I believe that increasingly it's going to be Sun's role and responsibility to gain the respect that it needs to get commit privileges on the software that it depends on for the systems that it's selling. So that's how I view our involvement with software and that's a big change from where we were 15 years ago. I mean I talk about this I talk about what's happening in software is a shift away from charging at the point of selecting the software to a point of charging at the point that the software begins to deliver value. And that's what I see Sun doing shifting from trying to charge for software based on a right to use license and shifting downstream to providing its customers with access to the people who can actually fix the bugs that are showing up in in production. Yeah I implement features and I do believe that works as a business model and we are already seeing it work as a business model on Solaris. Yeah but the the question for me is how how can we grow that as a so you know I have the luxury of being paid by a company to try and do good. They don't know that's what I'm trying to do that they think that what I'm that I'm actually working on the software and following the shareholders objectives but actually I'm paid to wander around trying to do good. So that that's that's you know it's a great job to have. I want to make sure that we do more and more good and that we do less and less unintentional evil. Yeah but preferably no intentional evil. Yeah I think it's also in the interest of the shareholders. Okay I think we should also try to kick off the discussion about the license CDDL. Okay which has been the biggest problem so far. So now I've got what I thought I'd do is ask Mark to moderate for us. Yeah because I tend to get deeply involved in talking about the license and I thought Mark could step in and kind of push people apart from each other and you know cool time out. I've got Don I've had lots of interesting discussions about this and I know he disagrees with me so it's unthinkable but who would have thought it. Also by a sheer stroke of luck we have got Denise in the room and she was the one that actually wrote the CDDL so she can answer any question about what's in there. I will say at the outset that I don't think the CDDL is a final license and that I'm happy to change the things that are broken on it if it makes sense but I will also say that the way the license is is that some of the things that are in there are there for very specific reasons. So for example the patent grant that comes with the source code is there to make sure that when you're part of the community of CDDL license software you can guarantee there's no one in that community that can make a patent attack on you and consequently I wouldn't want to take out the patent grant clause for that reason and I also would be very unhappy about dual licensing because the dual license would mean there was a way for people to be involved without granting rights to their patents unless the other license also resulted in a patent grant. So GPLV3 for example has got a patent grant built into it, meets all the objectives of CDDL but does it in a different social context so it might be a very appropriate dual license but as it stands for example the BSD license or the GPLV2 wouldn't really be appropriate dual licenses because they don't convey the patent protection that I want the community to have downstream. Even though there actually is an implicit patent grant in GPLV2 it's just not as explicit as it is in the current draft of E3. So we can have that discussion, let's see if I can get this up on the screen. Who's the video meister? Okay so while that's coming up let me first introduce myself I'm Mark Shuttleworth and I'm happy to play the role of semi-neutral, mostly neutral moderator in this discussion. Let me stay at the outset that the reason I'm interested in this is because I think there are two competing means, two competing ideas within the Debian community about the application and the use of Debian technology and you often see debates going on about something specific but what's going on underneath is this contest of rules between these two opposing ideas and those ideas are on the one hand kind of the narrow application of Debian and on the other hand the universal application of Debian and one of the things that I should state my bias in this, one of the things that that I really believe in and the things that I think Debian has done that's remarkable is that the meme of the universal application of the technology is a very powerful one and so one of the things that I think was at the heart of that huge flame war was this fight between people who want to see a sort of a narrow interpretation of what Debian is all about and where it gets used and people who want to see the general application, the universal application and I think that's kind of underlies this discussion and I certainly think that the most powerful idea is the universal approach, the idea of saying that this is good technology and we should spread this technology and we should look for ways to make this technology more widely available. Anyway so let me ask Don to take some minutes to sort of state one perspective and I'm sure we'll get some comments and feedback from that and then the discussion on the license can continue. Okay just first and foremost as always everything that I'm talking about is my own personal opinion not the opinion of Debian or anybody else if you think that it is there's something wrong with you not me. So first and foremost the current issues with the CDDL as I see it so there's actually been multiple versions of the CDDL that Debian legal itself has addressed. For the most part the license has been steadily improving and becoming closer and closer to what I like to call a non-controversially DFSG free license and so what I mean by that is that it both appeases the hardliners and the people who really don't particularly care what the license actually says so long as it kind of seems free. So I mean to get an idea at one end aspect you have some of the hardcore Debian legal contributors at the other end you have what tends to be known as the OSI and their interpretation of the open source definition. So the primary issues that legal has identified with the CDDL as it stands currently are both the choice of venue clause which requires that actions be brought in a specific venue which surprisingly actually isn't defined in this particular license. It's sort of vague as to what venue is actually brought in there. Right but that means that you could quite easily have the same work that specifies venue in multiple different venues which would be rather unusual. So Debian legal has for the most part decided that these sorts of clauses infringe the user's inherent freedom that is that they should have suits that are brought against them decided in the most logical venue which for most of you and I would actually be the venue in which we reside. If you're a big corporation of course a multi national corporation then it's less of an issue you have huge legal teams but for the common developer it makes more sense to have this sort of suit be brought in a local court where you actually have attorneys that you can understand a law system that you understand. So we felt that giving away that right that's inherent in most legal systems was something that you should not have happened by chance when you're using a piece of software. So that was the most major issue that I remember. There were also a couple issues with patent laws and the exact setup of which patents were granted by the license but those were for the most part a little bit more minor. Of course if you want to know the full context of what was an issue with the license you really have to look through the Debian legal threads. I don't have them memorized of course. I've sort of fuddled my brain thinking about GPL v3 and some of the clauses in there. So forgive me if I've forgotten something that was really major. Yes. So the license the CDDL is inherently GPL incompatible because it actually has a clause which just like the GPL forces the license to impose no additional terms. And of course the GPL imposes terms beyond the CDDL. So because there's some sort of you know you may not impose more terms that are defined in this license. They're inherently incompatible unless you magically manage to get a another clause seven exception in GPL v3. So as it stands right now I mean my interpretation is that they're incompatible licenses. With that stop it going into Debian. So long as there's not the formation of derivative work where in your linking for example a CDDL work with a GPL work it shouldn't be a problem for bringing a work into Debian. We have a lot of things that are GPL incompatible. The most classic one for example is lib SSL. So most works that are under lib SSL and link with the GPL work. The GPL work has to have a linking exception in order to allow linking with lib SSL or somebody has to port it to use GNU TLS instead. I'd like to come back on the choice of a new issue. Isn't that United States only program because just like choice of low issues clauses are considered to be free because I mean venues courts consider choice of low clauses just moot in many countries. Why wouldn't they do the same with the choice of venue? It would seem the most logical for them. Okay well so the issue with that is that in order to get rid of a choice of venue the first thing that you have to do is say that the choice of venue is invalid then you can actually change where the venue is. So I mean if you're in a country that's enlightened and agrees that you shouldn't have to travel around the world in order to defend yourself against the suit then so much the better. But not everybody has countries that immediately throw out choice of venue clauses. For example the US doesn't necessarily throw out choice of venue clauses instantly. I mean otherwise there would be no argument to include them in the license in the first place. I mean the main reason they're in there is when you've got two corporations that are having a lawsuit with each other and the choice of venue clause is then applicable if the two corporations happen to both be present in the state that you've chosen as your venue. Because when you've got two corporations that are present in all of the states of the US if you didn't have a choice of venue clause in there the first thing that would happen in the lawsuit would be an argument between the two corporations about which court to hold the lawsuit in. With both corporations picking the court that was going to be finding their favor as the venue. And so the reason you have a choice of venue clause in there is so that you don't have that debate. You know if Sun decides to sue Microsoft and the contract says that the venue is the courts of Santa Clara then the venue is the courts of Santa Clara because both Microsoft and Sun trade in California and therefore it's appropriate to have that as the court. The general view by lots of people that I've spoken to is that choice of venue clauses essentially null and void when the parties involved don't do business in the in overlapping territories. So there's no way that a court in Scandinavia is going to force you to be extradited to California to go conduct the case in the courts of Santa Clara. The only way that the company that was using a license that said the choice of venue was California would be able to work would be by pursuing you in your own country. So there's a strong argument that I've heard that the choice of venue clause is it is essentially a no up unless you're dealing with entities that do business in a variety of states. Right. And that's the argument that usually gets raised against the argument that the venue choice of venue is not free. Is that in fact the venue causes traditionally a no up. But I mean my basic argument is that is why bother to have it as well. And so my basic position there is that when we revise the license I think it'll go because the reason it was in there was because it was part of the license that we were that we were modifying. Again Denise is itching to explain how this all happened. I know that I'm actually itching to explain. But when a corporation writes a license any license they start with some ideas about what they want the license to do. In the case of Cuddle it was written for open Celeris and the design principles were we wanted to be as much as possible like Mozilla because we would have used Mozilla except we thought it needed to be updated. Mozilla has a hard coded choice of venue and choice of law. We wanted Cuddle to be as reusable as possible. So we D hard coded it. If you know what I'm saying we made it user configurable. But in order to our goal was to get the Mozilla community interested in adopting Cuddle as an update and a revision of their license and so we had to cover the bases that they were covering because you can't just do away with an entire piece of a license without the community wondering where it's going. Now in the case of the GPL there is no specific information about venue and choice of law because the GPL if you ask its author Evan Moglin is not in fact a license or a contract it doesn't have the same features that a traditional contract has. He thinks of it as a copyright notice and therefore it doesn't have to stipulate choice of law venue all that stuff. And it's almost like you guys are talking about two fruits that look similar but aren't really similar. Most of the things that are wrong with the Cuddle we knew about when we published it and the reason that it got published when it did had to do with when they were making their announcements about open slairs and also how far we could push the legal team in the time that we had allotted. But we actually had a set of errata that we wanted to see attacked in the next revision and Simon has very capably carried that work forward. In the case of the choice of venue if one of the balancing features is whether or not the Mozilla community is willing to see that go away because one of the features of the Cuddle was to try to do away with license proliferation. Many many many Mozilla based licenses exist. We were trying to roll a single tar ball if you will that encompassed all of their issues elegantly in a way that could be reused so that people would stop writing boutique licenses on the Mozilla model and start to use a single model. And that was a more important design point at the moment we were writing it than whether or not the W community liked it. It's Simon's enviable or unenviable position to take that work and make it work going forward. Yeah. But just so you guys understand that's why it's there. That's where it came from. There's no nefarious plot in there. It's literally just trying to cover the bases in Mozilla. By the way Mozilla was selected partially because it is GPL incompatible. That was part of the design when they released open Solaris those developers. There were people in that effort that were really hard pushing for just a wide open spaces license put it under BSD even put it under GPL. There were people who wanted to see that happen. And we couldn't convince the actual engineers who wrote open Solaris or wrote the Solaris Colonel to me this is a compelling argument. You guys are all hackers. You decided you wanted to work on a project that was licensed under the GPL. These guys worked for 15 years on a code base that was being open for the first time and they had some biases about how they wanted to see it opened. And you have to respect that because they're the engineers that wrote it. Not the company. The company wanted to see it go a little further than the engineers did. So anyway it's an interesting history and some day we'll write all about it. But having said that it's an educational process as well that so those those engineers are gradually learning what it's like to be out in the open. So that raises a couple of interesting questions. One Simon can you fork something that's under cuddle and change the change the choice of venue. You can't do that because the original owner of the code is the one that determines the choice of venue. So for Solaris for example the choice of law is the law of the state of California and the choice of venue is the courts of Santa Clara and that is fixed for all time. Nobody could could fork the open Solaris code and change the law and venue. So for all intents and purposes it's a the discussion that's happened on Debian Legal is a no op again when it comes to open Solaris because actually all the variables that people are arguing about are not variable they're actually constants. They're variable for the abstract license used abstractly for a theoretical project in the future. But for any given individual project they will be fixed by the original author of the code. Right. But that's one of the reasons why Debian Legal really doesn't like to talk about license in vacu. We like to talk them in the context of a specific application that's being brought up. And so that's one of the reasons to why the all the discussions about the CDDL most commentaries on it have the caveat that this is just a generic commentary about the license. Any work that uses this has to be you know reexamined and thought about as to how this all works out. And also the primary issues with the choice of law were really for ridiculous jurisdictions. Of course you are sorry ridiculous law systems. Obviously you may think that the U.S. is a ridiculous law system. But we were primarily thinking about law systems where you know things that would normally be allowed were totally disallowed. A great way to think about choice of law is it's like a namespace definition. It isn't about law actually it's about defining the words that are used for in the license. So the choice of law says whenever you run into a word you don't understand here is how to interpret it. It's the way that someone would interpret it if they were in a court in Santa Clara in California. And that's what the choice of the choice of law clause means. I actually think it's extremely reasonable. I also think it brings licensing certainty. I think not having a choice of law clause opens you up to the possibility that the law is going to be interpreted differently in different jurisdictions around the world. And personally I think a choice of law clause is essential in a license. So can you have a choice of law clause which which doesn't specify the venue of the case but does specify the interpretation of the of the language itself. Right and that's usually the distinction between choice of law and choice of venue. So legal for the most part assumes that choices of law reasonable choices of law are DFSG compliant. But choices of venue are probably not. Okay the second question that came up for me from Denise's comment was that this idea of deliberate incompatibility with a GPL. And there's this slightly scary concept of like poisoning eyeballs. Right the idea that if you've if you've gone and cast your eyeballs over code that's under one license that you then precluded from casting casting your eyeballs over code from another license. So so Simon would you comment on that and and and put the potential consequences of effectively making people choose where they where they can put their eyeballs effectively and dividing that the free software contributes a community. Well unfortunately that tainting thought is a is a is confusing two unrelated bits of law. Again so I've caveats here these are my opinions and I'm not a lawyer and I've never played one on TV. I've done so much licensing in the last couple of years that I've wondering wondering whether I could quit my job and go and do a law degree because I've just done so much of it of late but so what you're talking about there the tainting is to do with trade secrets it's not to do with copyright. The looking at somebody else's code and then going and implementing something which does the same thing is a problem if you are dealing with a trade secret it is it's actually not even a problem if you're dealing with a patented technology. It's only a problem if you're dealing with a trade secret and there's no there's really no problem whatever in looking at the code that's in open slurris and learning from it and going and re-implementing the same ideas in the Linux kernel for example. How do you how do secrets and public source code relate to that? They cannot be trade secrets in public source code. Right so fundamentally there can be nothing in open slurris which you could argue as a trade secret. So as soon as we licensed it under a an open source license as opposed to a free and open source license it was impossible for there to be trade secret cases anymore and so issues of tainting no longer apply. Right but you still have the issues of whether or not somebody is creating a derivative work based on the open slurris work if for example they've seen a particularly clever way of implementing something and it's not the only way it's original and then it sits in the back of their mind and then they regurgitate it later. I mean they're still creating a derivative work of something in open slurris so it's it's not as clear-cut as you can't you look at it and you're automatically tainted but it's a case of if you look at it you have to be aware of what you've looked at and not duplicating the same thing again. So this is this is a recipe for a scostar lawsuit right where someone claims that that a piece of Linux the Linux kernel bears a scale you know an unhealthy resemblance to something that was in open slurris. Again I'm really not allowed to comment on the scosu with the camera on but I I think they're on dirty ground just for the sake of argument. Can we switch the camera off? So the thought that you're pointing out there I've taken I've sought advice on that subject and of course the advice that I've received I've got to do the American caveat the advice that I've received is not applicable to you and you need to seek your own advice but the advice that I've received is that you would it would have to be pretty much cut and paste for there to be an issue. If you if you go and see how I have done my link lists and you go do link lists that work the same way if you haven't cut and paste the code it's almost certainly not a derivative work. Right yeah that's a question of the difference between the method and the implementation. Okay more comments from from from the floor can we do do we have any comments from folks who haven't spoken yet? We've got two more here from folks who have anybody else got a hand up? No you're closer. About choosing a derivatively choosing a license that is incompatible with the GPL isn't Sun shooting itself its own food because Sun is also contributing a lot to developing GNOME which has large parts license and the GPL it sounds rather veiled to then distribute a system which consists in pieces with two incompatible licenses. So there's a couple of different issues there. The first issue to point out is that as Denise pointed out the license that there is on open Solaris is there almost exclusively because the engineering community in Sun which is a big engineering community it's so forget this camera it's about that there's about a thousand people on that in that engineering community and the people in that engineering community fairly clearly expressed the view that they wanted to use a license like BSD and we didn't feel that BSD was a very good idea for them to use for that code for a variety of reasons. We wanted them to use a copy left license because we wanted to make sure that there was a there was a a republication of derived works. So then they say I actually disagree with Denise to a certain degree. I don't think there was any chance we could have used the GPL for open Solaris because I think that it would have meant that our release schedule would have moved out by several years because we would not have been able to continue. What we did when we launched open Solaris was most of the source code for operating system and networking which is what the Solaris folks called kernel was released but there were one or two parts where we weren't completely sure we had the right to re-license it and so we launched with those parts as just as binaries and we wouldn't have been able to do that under the GPL. We would have had to have waited until we had confident provenance on all the parts that you needed to link together the operating system. Right though that's actually not quite true because as the original copyright holder you can actually put in exceptions. We could have used exceptions but that would have then led to a very nasty inequity that we were the only people who could actually build the operating system. Right and also the only people who could distribute it as well. And we thought that was a really lousy thing to do. Now some things to note first of all Sun uses GPL v2 for other things so I'm very happy for us to for example license the looking glass code the 3D user interface code under GPL. I'm very happy that we've licensed the Ultraspark T1 very long code under GPL and I expect us to use GPL actually for more and more projects going forward in the future but this community said essentially the people who really matter in that community said you do that and we quit. So faced with that situation I really didn't think it was worth the extra three years delay and having all the people quit to be able to use the GPL. Now what practical problem does it cause which is the second part of the question. I actually don't think it causes very many practical problems because the place that the chief problem comes is if you're going to make derivative works that then combine together GPL licensed code and at the operating system and networking layer everything below there is all licensed in the same way. There isn't a need to intermingle GPL code in and if you were going to take ideas that were in the open Solaris operating system and networking you couldn't cut and paste them into your derivative work. So you're quite welcome to come across and take the code that does detrace or zones out of open Solaris and reimplement it in the Linux kernel. I don't think you'll have any trouble doing that but I don't think you'll be able to cut and paste the code because the threading model and the architectural model of the two kernels is so radically different that cut and paste isn't an option. So the practical barrier for that code and the influence it has on the new Linux community I think is exactly minimal. But in terms of getting the code as free software we were actually able to get it as free software and we were actually able to do it a year ago. If we hadn't picked the license that we did we would still be working out how to do it in two years time. Okay we had we had another comment of him. Yeah two other issues about choice of venue. If you take two programs that are licensed on the CDDL and have different venues aren't they incomparable at this point? That's a really good question in fact because the license itself specifies that you cannot impose any additional conditions and having two conflicting choices of venue would appear to me anyway a priori to be two conflicting conditions. It probably would. It hasn't happened so far because everyone has conveniently picked California and Santa Clara as their choice of law and choice of venue. But it's one really good reason why next time around I intend to get rid of both. Yeah I'm probably thinking of York Schilling who's realising all his stuff on the CDDL and I don't think he's also using Santa Clara as a choice of venue. Right yeah well that's a whole other issue as far as understanding of the license that is being licensed under and what it actually means for your work. And so unfortunately he has a history of not quite fully analysing the licenses that he's worked under and what it means for himself as well as the others who are using his code. Yeah and the other issue is that people are basically objecting about the Solaris Lip C which is under CDDL being linked against GPL programs. Okay. Carry on. That was basically the main objection I've heard about against the CDDL because you basically would link GNOME under Solaris against the CDDL license Lip C which wouldn't be allowed. Right and exactly and the typical way that this is explained away is by trying to pull in the what's colloquially called the operating system exception in GPL version 2. And so that applies when you have separate entities distributing the operating system. Okay. So that kernel compiler Lip C stuff like that. And then a separate entity, a wholly separate entity distributing the GPL stuff like GNOME and whatever else that's GPL. Then the OS exception applies. Under some interpretations. Well yeah obviously it's English. So the again the advice that I've received that you can't rely on is that the operating the operating system exception is most likely to be read in a way that does not cause an issue with the use of either Solaris Lip C for GPL license code or the GCC Lip C with CDDL license code. And I don't know whether it's even worth trying to do it but I have a video clip here from the GPL v3 conference in Boston where someone actually asked Evan Mogul on this question. And Evan recognizes that there are interpretation issues of the operating system exception in GPL v2. And should we play that because we're just about time to wrap up. So let's get that. We're going to use some high tech to do this. Number. I have a number of developers who work on GCC. And in one of the patent slides I don't remember which one you spoke about interfacing with patent covered libraries that may be part of an operating system that they may end up interfacing with. So what I'm worried about is it might be demoralizing a bit to know that their changes to GCC may not be able to be run on Windows or Sun operating systems even though we don't rely on those systems. On the contrary one of the reasons that we altered what had been called the system library exception was that questions had been raised about precisely that sort of case. Was it inappropriate to link GCC with a C library under proprietary terms or a C library under free but GPL in compatible terms. The language of GPL to seem to us to state that that was permissible. But there was legitimate reason for doubt and we wanted to make it clear that we always thought that mode of employee was reasonable and we always thought it should be permitted and it is by this language now explicitly permitted. We hope on the contrary that rather than creating moral difficulties for those development projects we should be able to give them better legal assurance that what they are doing is fully permissible and fully compliant. So he spoke there about linking. Did he address distribution? Yeah, and that's the critical point because there's a serious difference between setting up and distributing just a part that happens to use an operating system. For example distributing a GPL to work which happens to link with the window DLL or something which obviously isn't available under GPL compatible terms and then turning around and distributing both windows with the GPL to work. So as I say the general interpretation that I've been I've been asking around trying to understand what this question boils down to and the answer is that if you're using GPL v2 there are perfectly reasonable perfectly reasonable cases that can be made both for it to be perfectly acceptable to link at the libc and for it to be unacceptable to link the libc and ultimately which of those was true would depend on what happened in court when you tried to actually impose the terms. Ultimately the question of which of those you're going to accept turns out in my view to be a social issue rather than a legal issue because there are two valid and defensible interpretations which a skilled orator would be able to persuade you. Two skilled orators will be able to persuade you they were right arguing both ways and ultimately you have to decide socially which one is the one that you want to accept. But the v3 I'm just trying to drive to a conclusion. The v3 license makes this all go away right. There is explicit language. Again my understanding of the current draft one of GPL v3 is that this issue is fully articulated and doesn't occur any longer. However what will be in draft two and what will turn up in the final GPL three isn't certain. But it certainly seems to be both Evans and Richard's intent that there shouldn't be dispute over this. That when you've got two different communities who are developing free software they should be able to work together. They shouldn't be artificially divided by optional interpretations of the license. Because most of us are more familiar with the Linux kernel community do you want to take a minute and just talk a little bit about the open Solaris community? Just for interest sake. I can do. So what can I tell you about the open Solaris community? What's the community experience like and how's the community developing? So the open Solaris community started out last June by getting a whole load of Solaris developers from inside Sun and getting them to open up their number to accept contributions from anyone that wanted to make contributions. The logistics of setting that up have turned out to be fascinatingly difficult. So because we were using a very elderly version control system that was the predecessor to BitKeeper, it was actually written by the author of BitKeeper before he went and got a job. And actually he was working for Sun before he started his own company. And disentangling ourselves from that distributed version control system has turned out to take a lot longer than we expected. So we don't have a public VCS at the moment. So when you make contributions on open Solaris, what you do is you send an email to a list and get a sponsor to help make the contribution for you for the VCS as it stands at the moment. At the moment there are nearly 14,000 people who are participating in the open Solaris community. I'm guessing there were around about 1,500 that are taking the source code and doing things with it regularly. So far we've had about 100 commits made from community contributions back to the source code. Over time, there's going to be a big jump when we get Subversion and Mercurial, which are our new VCS selection, sorry Mark, to be deployed out on the website. As soon as that happens, then I'm expecting we'll see a whole load of people will be given direct commit privileges on the source code base rather than having to work constantly through sponsors. At the moment we've also seen five different distributions being made from the source code. There is, of course, the now famous Nex Enter OS is one of them. There are two command prompt-only distributions, Shillix made by the renowned George Shilling, and one made by, called Belenix, made by a group of engineers in India. And there is a spark distribution as well called Martucks. The community is continuing to grow pretty steadily with new interested parties coming in, with people doing things like there's a group who are porting the Solaris kernel to PowerPC and doing pretty well at it. There, what other random projects are going on? I think that's a good flavor. It's a growing community. It's a growing community. It's also a peer community. The way that Solaris worked inside Sun was it was a big developer community that made its own decisions. So Solaris was never made by having some executive up here saying, we will have this feature. Solaris has always worked as a college of engineers who are making decisions within their college. And all that switching to open Solaris did was the doors of that community are now open. It actually wasn't a great big leap to become a free software community for them. They were operating as a free software community effectively inside Sun. Now they're an actual free software community that has got open membership and forkable source code and a free software license. OK, we are out of time. Let me say that one of the things that I think Debian does a couple of things better than any other organization institution in the world. And one of those things is actually Debian Legal because the fact that you have intense discussions about things that many other people would just, you know, not want to bother with too deeply makes this a fairly unique contribution to the whole process of the free software community evolution. So thank you very much. And thank you, guys. Thank you for not throwing rocks. I appreciate it.