 Good morning. Welcome to Keith Packard's ex-community history talk. Thank you very much Thank you for the wonderful introduction as always This morning's talk is completely bling-free. I Know all of you come to watch the bling, but I presented absolutely no interesting effects on the screen all week Spent the morning packaging free type I'm gonna talk about how we came to be how the ex-community came to be where it is today in an effort to try to explain what happens through the unintended consequences of rash and Hurried action. Thank you. That was rash. Well, let's see if it comes back Okay, okay, I wanted to talk about where X came from X of course came from originally came from Stanford the The standard V group who are building the V kernel, which is a distributed multiprocessing system across the network Built a graphic system called VGTS Policente eventually turned that into a more formal windowing system called W They had the V kernel and the W windowing system Eventually that went over to MIT and Bob Schachtler played with that and got it working on Unix and eventually changed it enough That he called it X Through the air from X one through X ten it was a kind of a typical university project There wasn't any particular formal schedule There wasn't any particular formal requirements for compatibility or anything like that every time the protocol changed in an incompatible way We got a new version number One of these numbers doesn't actually exist in reality X nine was never it was never actually instantiated Each version of the window system was fixed there weren't any it wasn't extensible in any way And so every time they did want to add some new feature like color They had to change the protocol number In about 1986 1987 a bunch of Unix developers a Unix systems vendors Were starting to ship X 10 and they discovered that X 10 really wasn't going to do what they wanted to do But they wanted something very much like it So instead of starting from scratch with an entirely new group of people they said well what what happens if we just You know help the development of this project and come up with a new version that's going to be have a little longer lifetime But they succeeded beyond their wildest imaginations in that regard So they actually complete they started completely over they built a team of Engineers that designed the protocol digital actually donated an entire staff of people to implement an entirely new X X server and X library It was a really wonderful industry. Oh Yeah, I forgot to reset my screensaver. I'm sorry I was a very really wonderful example of industry university collaboration and in kind of its best form the industry provided a lot of manpower The University provided a lot of intellect and together they worked together quite wonderfully He's laughing at me We actually at universities a lot of you know university participation in this they are we had Stanford we had CMU We had Purdue we had a lot of other universities involved in the window system at this point There were even community people outside of the university system involved although they were a lot rarer It was very hard to get an internet connection in this era without university or industry collaboration connection Now one of the interesting things that happened during x development, which is very relevant to what happened politically Is was the collapse of the internet in 1987? How many of you were using the internet in 1987? Yeah, that would be B Dale and I What happened was this is pre the pre Van Jacobson slow start TCP implementation in this era Any packet loss if you lost a packet in the middle of a packet stream The sender would retransmit every packet that it had that it hadn't been act that had not been act yet Which is to say if you lose one packet in the middle of your window you'd retransmit the entire window Which is pretty exciting because that meant as soon as the network started to get a little congested And you started to lose packets you started to flood the network with packets Anybody see a kind of a the wrong kind of feedback going on here? Yeah, it's a positive feedback loop And so as soon as the network started to slow down during the particular day it would stop dead There would be effectively no you you usable internet What x was the x11 development was going on in the middle of this nightmare and what happened was They stopped being able to communicate between the two major centers development at digital research I mean at digital Western research lab doing the implementation at an MIT doing the specification and Implementation of the library so they all of a sudden their network died Can you imagine what would happen to Debian if the internet suddenly failed? Right here we are sitting around you're actually listening to me why Because there's no network Yeah, so the x developers were in a quandary what the heck to do well they started Doing the classic days of your time sharing stuff They would start doing FTPs at three in the morning Fortunately was the US the internet was largely US based at that point So there really was a notion the diurnal cycle of internet usage So you really could take advantage of that and use the internet at three in the morning And of course my screensaver fires again. Sorry about that The other thing is started to do was sneaker net right stick a mag tape of source in a in in a FedEx envelope and ship it across the country Fortunately FedEx still existed and it wasn't dependent upon the internet like it is today Okay, so how did they respond to the failure of the internet well they responded in some very bad ways They didn't understand that it was a short-term failure in even though van Jacobsen had already demonstrated a successful internet Inter operation with it or they certainly changed tcp implementation So what they did was they said well we can't trust the internet for our development Even though we've successfully developed x11 in a distributed fashion largely using the internet We don't believe that we can do ongoing x development trusting the internet and doing collaborative distributed development So they gave up They gave up by centralizing the development of the x-windows system at MIT in hiring a bunch of staff Well, what happens when you hire a bunch of people to do development? Well it requires a bunch of money Which means you need to go out and collect money from industry people because academics don't have money to donate to this sort of thing And you have to find developers who are now Central up who are now the privileged developers in that community right you have people who are paid to work full-time on the system And who are located where the source code lives and who grant access grudgingly to people outside of their cloister So we move totally to a cathedral model at this point Of course the consequence of moving to a cathedral model was that this consortium became addicted to money Right we had between four and seven developers do the math right it took a huge amount of money to run this organization As a result the consortium needed to continue to find sources for funding itself And so it needed to find ways to incent corporations to become members Those corporations were granted special privileges one of which was access to the source code during the development process not at a release cycle So the people outside of the corporate world only saw acts as these point releases They didn't see the development process. They couldn't get access to the source code repository They couldn't contribute to voting on what parts of the system would become the standard So all of a sudden we had no collaboration from outside of industry in fact all of our Academic collaboration disappeared at the same time Kind of a catastrophe are our community collapsed from this broad-spread Collection of collaborators from universities and outside of universities and industry from this partnership All of a sudden it became a bunch of contract programmers working in MIT for corporate interests Does that sound like an open-source project to you? It didn't feel like it to me at the time But I was young and naive What did I know? At about the same time about 1991 or so this university student in in Germany Thomas Rohl Was writing X drivers for VGA cards? Of course PCs being pieces of crap that they were at the time The X consortium had really no interest in them, right? They were totally uninteresting pieces of hardware. They were not the Unix vendors workstation products They were you know machines sold for multiple thousands of dollars instead of multiple tens of thousands of dollars So the X consortium as a body was uninterested and that trickled down into the staff of the X consortium in MIT Where I was at the time and we were completely I mean we just laughed at them We said ha ha trying to run X on a toy like that. How funny? But Thomas was actually very interested in in putting together and working on this stuff You've been doing a lot of stuff on it was a pretty cool project Was really interesting that you could take a $2,000 machine at the time and get a full-fledged Unix with a window system running on it Was pretty interesting So what Thomas does is he joins up with a man card called Mark Snidley and they put together a consulting company SGCS and in order to get their code this VGA drivers into the window system They join the consortium the only way that you can get a voice in the X window system is to pay money It's a totally pay-for-say environment As a result because they were paying money suddenly they had a voice and they could get their code shipped in a free software project How many of you pay $10,000 a year to get your code shipped in a free software project? Yeah, yeah like it had happened today And it actually did work X3d6 code was included in X11R5. It was very early. It was very primitive It wasn't actually all that useful at the time But of course Snidley graphics consulting services had this monkey on their back $10,000 a year in order to get their code into the window system Because they were paying money for it. They thought there was actual value there so they put together a Commercial product called XIX inside which many of you may have heard of it's a commercial X server for PCs running Unix To to kind of justify their existence right they had to put together this commercial So Thomas started out as a university student doing free software on a spare time and he gets sucked into the whole corporate Corporate world and dragged down into this commercial X server vendor and Xi's you know always eat down a kind of little marginal existence doing X servers and they spent a long time in the late Night. He's actually bashing the free implementation for being slow and not very well supported You know not exactly a great collaborative environment Not a lot of friends were made during that time between the x3d6 people and the X side people Also, where did x3d6 come from? Haha. Well, exactly the same story repeats itself There's a group of people wanting to improve these barely usable X drivers for PCs And you know, they're regular open-source developers. They're just having a good time developing free software for their hardware They're scratching their own itch they get together and start collaborating on patches for x11r5. That's x3d6 drivers But they become very frustrated by Thomas and Mark's nidalee's Company who are not now contributing their patches back to the back to the X consortium They've taken all of their x3d6 code and they made it all closed source the MIT license allows them to do that But so the x3d6 people were not able to work with X Xi and Collaborate collaborate in the development of this stuff at that point. So They put together a bunch of patches. They actually released a version of x3d6 x3d6 1.2e or something and Build the first free software version of a largely workable PC PC Unix-based Windows system Xi and X and Thomas roll and Mark's nidalee say wait a minute x3d6. That's our trademark and So x3d6 was actually forced to change the name of their of what they were shipping And so they chose kind of a pun on x3d6 and called it x3d6 noting that it was now a freely available with distribution Probably one of the you know one of the early early notions of free software as being a distinct product from From a commercial variant that provides the same functionality Of course x3d6 is now a bunch of free software developers who have no money And so what does the X consortium do? They are ignored And again in order to try to get their work recognized by the X consortium and included in X releases What do they do? They put together a corporation collect money and join the X consortium But unfortunately because they are not a credible commercial company. They've made their intentions clear They're just going to be doing free software This corporate consortium Continues to ignore them even though they're paying money now. So it's not enough to pay the man You have to pay the man and become the man you have to play by the rules You have to pretend you're a corporation and pretend you're interested in making piles of money from free software So you can imagine how the x3d6 developers felt about this great development, right? They were they were a happy bunch of campers The x3d6 project was focused on PC graphics drivers. That was really the work They were doing they were taking the X window system and they were porting it to free to PC Unix systems Of course when they started Linux didn't even exist And so they supported a wide range of operating systems both commercial and free operating systems It's important to note that they started with commercial operating systems. So they had no access to the kernel development process Commercial Unix systems didn't offer development kits to random developers because after all they could get money for that So the 3d6 developers x3d6 developers couldn't actually change the operating system Which is why you see the X window system architected the way it is today The system that we run today doesn't depend upon the kernel for anything, right? It has its own PCI bus remapping logic It has its own, you know mode setting logic outside of user mode All that requires from the kernel is to be able to map devices and execute IO IO operations that doesn't require anything else from the kernel If the project was started today, nobody would even ever consider doing it this way So when you laugh at the X window system and you say why does x3d6 or Zorg at this point do all this stuff in user mode It's because of this origin the x3d6 developers were free software developers working in a commercial world And they didn't they couldn't play by the right they couldn't play by the commercial rules They didn't have the money They didn't have the resources and they wanted to be able to release their code for free The developers for x3d6 were almost entirely uninterested in issues beyond the driver environment Right, they were they were happy little hardware hackers writing drivers for ET4000 cards or whatever they had at the time So they were not really all too concerned about toolkits and libraries and network protocols and 3d systems and all this kind of stuff They really were focused down at the low level and it's important to remember their interests as we move along in history So what happened in about 1996 was The X consortium finally shut down the commercial UNIX vendor said oh wait a minute. We're not making any money in this environment anymore Why are we supporting a consortium for this? We really don't want to pay money for this So they all kind of walked away. They said well, you know X is just gonna die You know it's gonna it's gonna disappear because it's a UNIX based technology and we all know the world was running to windows But somebody needed to hold the copyrights so as usual and it came out of a casual decision the X Consortium Organizers said well, we'll just pass off the copyrights to the code To the open group and let them you know continue to support it so that our existing legacy dying products I will have some kind of tail of support as they exit exit the market Well the open group grabbed on X and said oh wait a minute these UNIX vendors have to ship it It's part of the UNIX standard Right if you want to get labeled with Tog as a UNIX system You actually have to support and you wanted to support X you have to actually have to cooperate with Tog Because they own the trade the copyrights now right can't ship the code without the cooperation So what Tog decides to do is Tog says, you know, we need to support the system for these legacy UNIX vendors We don't want to do it from for gratis. So we're going to try to extract money from X So their big plan was to re-license X in such a way that you couldn't have a free software implementation You actually had to pay for it again because they had to pay staff to maintain the Windows system They had to have some revenue stream. It wasn't entirely evil And so they attempt to re-license X X11R6 was going to come out with a closed-source license that required you to pay Money to Tog to ship ship products compatible with it Fortunately X was had been released under the MIT license until then so x3d6 said well if you ship a closed-source version We will support the old code and not include any of your patches as free software, you know great instance of where a free software group was able to successfully force the issue and Tog basically said, okay, you know, we'll we'll re-license it But the effect of this was that political control the X Windows system was very clearly now had moved from these commercial companies Into the free software community the free the commercial companies no longer had any political leverage to force license changes in the Windows system They had no leverage change how the Windows system operated And it was very clear at this point that the free software community had won and that x3d6 was now in de facto control of the Windows systems future The problem was what are the core x3d6 developers want? They want to build drivers. They don't want to own the Windows system They're kind of frustrated that this whole notion that all of a sudden this huge ball of code is now Landed in their laps. We have a question over here. I have one question. How long did the X consortium had its own developers? The X consortium had its own developers from 1987 late 1987 all the way till 1996 so about nine years I was there from 1988 till 1992 It moved out of MIT in 1993 and became a separate organization But continued to have their own developers until 1997 in 1996 at which point it was moved to Tog And that some a couple of the developers from the X consortium moved to Tog as well to continue supporting X Windows system there So Tog has actually had a very small number of developers I don't know exactly when they stopped having any developers working on it. Maybe 1990 late 1997 so okay So now we have the free software community in charge of X We have these legacy Unix vendors still shipping ancient crafty workstations running X and we have a bunch of PCX server vendors selling X in the Windows market The the Unix vendors still need X. They actually still there is still a market for workstations running Unix That include the X Windows system. They can't just drop support for it But their market shares not their market is not expanding radically, right? Anybody recently bought a HP UX workstation? Yeah, none of us So they but they needed some way to yeah pdl says, okay, so that market's not doing so good It's still a credible market, right? As long as you don't spend too much money people are paying a huge premium for Unix workstations because they're way more reliable than our PCs They're way they were until recently way more powerful and they had tremendous vendor support, you know stability Supportability expandability all kinds of cool features about Unix workstations that made them valuable in certain markets and continue to make them valuable in some markets So these Unix vendors are making, you know, 30 or 40 percent margins on these products They don't want to just throw their products in the trash, you know, they're having fewer and fewer customers every year So they still need X. They still need to do some support So the Unix vendors get put together Kind of get together and say well we need somebody to support X so they basically They force Tog because they're still related to Tog to slightly dissociate the X portion of Tog from Tog So they have a little more control over it And then they hire a consulting company to do patches so they can track bugs and fix stuff in a collaborative environment Classic industry consortium. There's a shared resource. They're each putting money into the pile So they can each benefit and they don't have to pay all the bug fixing separately It's still industry consortium. You pay if you want your bugs fixed. It's not too exciting. Most of the bug fixes come from where? X3d6, that's right So the free software developers are now sharing their work with commercial vendors Commercial Unix vendors not a lot of quite. It's basically a one-way street. Our bug fixes happen In the X3d6 world and they trickle down into the X.org environment and the commercial Unix vendors get their bugs fixed Basically by hiring the consultant to apply X3d6 patches. They're happy with the organ with a setup There's a little political wrangling and some financial machinations going on again because there's enough money there that there's interest from From third parties to try to extract money from the organization So there's nasty politics going on, but it's working surprisingly well for them But it's not doing exactly what they want Okay, so what are the unintended consequences? Well, what happened here? I lose a slide No, apparently not. Okay. Well, we'll just make do with what the contents of the slides say because after all the slide has to Direct the kinds of conversation here. I Think I did the talk yesterday almost entirely out slides worked a lot better Okay, so So the commercial Unix vendors are pretty happy with their little consortium But the problem is is this these commercial Unix vendors are starting to enter the PC market They're starting to migrate their hardware from custom graphics hardware to commodity hardware They still want to be able to ship their commercial Unix versions on this on the commodity hardware, but they're trying to get out of the hardware business HP and Sun and a deck at the time are starting to get out of the heart the commercial graphics card business out of the out of the Commercial boss of business. So they're moving this industry standard architectures They're moving the industry standard backplains and that kind of stuff And so all of a sudden their hardware is moving from stuff that they have control over that's moving really slowly Harder that's moving very fast. All of a sudden they aren't able to keep up with the developments and the graphics drivers They need who has drivers for this commodity hardware X3d6 has drivers for this commodity hardware So all of a sudden these commercial Unix vendors have a much more desperate need to be able to track Developments in the X world at a faster rate because the hardware is moving too fast for them to keep up with on their own So they're interested in collaborating more closely with X3d6 to get X3d6 code into their commercial Unix distributions to support this new hardware X3d6 meanwhile is has has starting to grow as well X3d6 is getting a huge influx of new interest and new members who are interested in working beyond the driver level I joined X3d6 in 1999 and I'm totally uninterested in graphics drivers I work slightly above that level doing graphics infrastructure and system system engineering There's other other people working in font design. We had a huge influx of Unicode support at the in this time From some people in the UK Mark Leisher and a bunch of other people at the University of New Mexico as well Who are interested in getting a broader internationalization support into the Windows system? So they're working way up in user mode doing stuff to Xlib to make it support UTF-8 and other Unicode stuff And there's people working in X-Term supporting X-Term to make it do all kinds of crazy new stuff There's people working all over the Windows system. The problem is is that the membership of the X3d6 corporation Remember, why did they create this corporation? They created the corporation to join the X consortium that hopes the X consortium would take their code and ship at which they never did So they put the corporation together quickly It's a Texas corporation who would organize a US corporation in Texas unless they were not interested in long-term Viability, right? It's not a very friendly corporate environment in a lot of ways They put it together with the minimal by-laws. They could just satisfy the requirements of the state of Texas And the state of Texas required that they have a membership with officers in an election process So many you know lots of organizations that we know have membership in officers and election process The problem was the state of Texas didn't require that they be able to elect new members so their corporate charter had a fixed membership not unusual for a Corporation to have membership that is fixed by the people who put the money in well There's not a lot of money here, but the membership of X3d6 was fixed All but one of these members the original founders of X3d6 leave the X3d6 project and go off and do other stuff They are still members of the corporation and in fact, they still run the corporation today But they have no interest in X in fact one of them is working at AOL doing Windows stuff One of them is off working on telecom stuff One of them is back to his law practice doing law stuff. They don't have any interest in what's going on in X They just want to basically be left alone the remaining member the only remaining member of Of the X3d6 project becomes a de facto dictator He has the only voice in the community which has any authority over the from the X3d6 from a corporate perspective So in order to say what X3d6 wants there's only one person with a voice there The chart of the organization again put together quickly to satisfy the needs of the state of Texas So they become a corporation so they could join the X consortium so they can get the X consortium to distribute their code There's no process in the Constitution of this corporation for fixing the governance model I suggested on a couple emails that maybe they should consider rewriting our Constitution so that we could fix the governance model But they kind of laughed at me so what happens is Because there's no There's no voice for the community within this project The developers became very unhappy with the direction the leadership was taking the leadership was shutting down development a lot of areas It was not interested in taking over its assumed role. What happened in 1999 X3d6 takes over the entire X window system But the core developers of the project and in particular the project leader really is not interested in things beyond the driver level So he has no interest in managing this very large project I mean the X window system is you know several million lines of code of which the core developers were interested in you know 10 or 20,000 lines So the developers outside of this area became very unhappy with the leadership there were changes that were not getting done There was no there was no say in how the project was run and in fact Because the core developers for the project had had such a bad experience with the X consortium I mean here they're paying their money. They're trying to play by the rules and the corporations laugh at them the Core members because how very distrustful of corporate involvement in the window system again And most of these new developers are being sponsored by corporations who are interested in free software development of the window system I was working for Suze at the time and then HP other developers who were working for Sun or for Red Hat So the key developers of X3d6 were very distrustful this they were concerned that if they granted any sort of control To corporate sponsored developers the corporations would again take over the window system and they would lose their window system again and have to do bad things so instead of instead of Allowing these new members who have are sponsored by corporations to gain some measure of control the project They kicked the members out of the project. They said you can no longer commit to the project You're not welcome to play here. And so we left And at about the same time Well, the this core developer the key developers was interested in ensuring that his own name Remain associated with the project because we're not for a long time and worked on it for almost 10 years So he assigned a new license to pieces of the project that were not a non-GPL compatible license that had essentially an advertising clause Just like the BST license slightly different advertising clause But it was clearly not GPL compatible when the FSF even put together a little statement that said this is not a free GPL compatible license So at that point we had a group of developers the bulk of the developers are no longer Franchised within the X3d6 project. We had a credible political reason for no longer shipping X3d6 We had basically quorum and a credible reason for for forking the project. So we did So what we did in fact was we hijacked the X.org the X.org name right the Linux developers the free software developers working in with a window system got together with the Small group of Unix system vendors who are still working in X.org and said what if we got together because we have shared common interest and Put together a free software project, you know reconstituted the organization with a legitimate license with I mean with the legitimate Constitution put together an organization that was run by free software developers for the benefit of the X window system and implementation and Basically created an organization from whole cloth We use the same name or a slight variant on the name It used to be the X org the X org foundation consortium. It was an industry consortium. It's now a been totally reconstituted as a non-profit educational foundation So we can take charitable contributions We have very strict rules because it's a charitable organization about how we can spend the money and who and where the money can come from And who is in charge of the organization in particular We no longer have corporate control over the politics the organization The governance of X org is entirely elected from within the developer community It's not elected from within the corporate controllers of the organization The money has been entirely separated from the development process You no longer get the sponsors of the organization no longer have any say in the development of the project The only thing that they can do is if they don't like what the project's doing they can stop paying And at this point with over a quarter million dollars in the bank It doesn't really worry me too much if they all left because X org is not exactly an expensive organization to run So how is X org put together? membership is based on work You put together a little statement of what you've done for X recently or what you plan to do for X and You send it in the X org membership board and the membership board says oh, yeah, this guy's worked on X You know he should be a member. What is membership good for the only thing membership is good for is voting for the Governance of the organization. There's no there's no tie between membership and development That we have many developers that work for their work on X that have no membership It's a little frustrating to me because I would like them all to be members So they would be able to vote in the in the governing Governing elections, but we decided not to tie those two together because we really wanted to make sure That development wasn't tied to the corporate structure in any way So we have you know dozens and dozens of committers Who have no interest in the in the governance of the other of the organization, which is like I say frustrating to me And I'd like to be able to fix that The board is elected by these members and that's the only thing membership buys you is the ability to elect the board The sponsors the organization are corporations who think that X X org is doing a good job and they're willing to put in a little money every year to help the organization thrive What is the money good? What is the money used for well right now? It's sitting in a bank account We haven't really figured out exactly what to do with the money It's growing. We're getting more of it every year, which is kind of cool But we have not figured out what to do with these resources and the reason for that is that the sponsors our Corporations interested in commercializing the X window system and X org foundation is a bunch of free software Developers interested in spreading their software through the world for free So you see right now a tension between The free software developers who are running the organization and the sponsors who are putting money in the sponsors are entirely Corporations interested in commercializing the window system now So far we haven't really found a good balance of how to make those two groups Actually figure out what to do with the money that's pouring into the organization The board actually controls the bank account on the treasure of the organization The board elects on what we spend money on and I and one or two other people sign the checks So the sponsors can't spend money the only people who can spend money is the board However, we acknowledge that the sponsors are providing the money and if we spend money in the way the sponsors don't like They'll probably stop providing money So we're trying to figure out how to do things that the sponsors like that that are useful for the organization And so far we haven't found anything that the sponsors think we should do we aren't going to do advertising We aren't going to do we aren't going to spend a bunch of a bunch of money on paying people to write code because we've discovered That's a really bad plan. We aren't going to spend money You know so far the sponsors aren't interested in spending money to get the developers together in any sort of massive way You know, it seems like one of the key things that a free software project can do is to bring developers together Right. Why don't we all come to deb con we come to deb con because it's fun to see our friends and because we connect in a Personal way which really helps the project move forward in a technical way so far We haven't convinced the sponsors. This would be a good plan for them to spend their money on so maybe we will I don't know So far we spent like maybe ten thousand dollars in the last two years. I could check the checkbook, but yeah, it's kind of amazing So I have piles of money nothing to do with it. Okay, so What's the result of this grand glorious political adventure? We have a lot of new contributors. We have a tremendous amount of new interest and Progress being made in the window system a lot and things are expanding a lot of different fronts one of the obvious advantages of having of letting pretty much anybody who has a cool idea into the pool of people who are Organizing organizing what gets shipped is that you get interesting new technology and sometimes you get multiple implementations of similar technology So right now we have both the XGL and AI GLX in the source code and people are experimenting with these two assistants Try to figure out. What's the right way to move the window system forward? We have changed the release process from everything used to be that everything that was in the Everything that was in the source code system got shipped, you know, you basically if you wanted if you If you were willing to accept source code maintenance for the project you were committing to ship the project and we figured out that that was probably not a great plan because there's a huge amount of New technology and a huge amount of competing projects And it's it's important to provide hosting for them and to provide a put place for them to collaborate But it's not required that we commit to that being a part of the window system implementation as a standard Right, so we're able to ship things like XGL and AI GLX and say yeah So those are competing implementations of a similar idea. We don't really say either one of those as standard So in in there's a whole bunch of other code going into the repository right now that we aren't even going to ship at all It's just sitting in the repository the release manager says oh that stuff's not quite ready to ship yet And we don't ship it, but it's in the repository so people can see it We have a shared collaboration environment that doesn't demand that everything in that environment be part of the release There are major structural changes, of course that have under that have happened in the last couple years in terms of development model and Politics the main thing that I like about it is it's a nice place to play now X3d6 used to be kind of a hostile environment because there was this you know There's this very clear hierarchy of the the key developer followed by kind of a cabal of central Central committers followed by the plebeian committers who had no access to the repository We've flattened it out now so that anybody who has credible credible Ideas is allowed to commit directly to the repository and get their changes in without Significantly without significant roadblocks in their way and that's really helped it become a lot more friendly place to play It's like well sure if you want to do some work go ahead Here's commit access to the repository have a good time. Let us know if you have any questions Okay, so what lessons did we learn well? We learned at the start of the project that quick fixes to problems to temporary problems have very long-term consequences the the foundation of the ex insorcium was Predicated on the lack of a viable internet, so here we have 20 years of political history that basically Caused by a short-term technological failure right x3d6 and GCS those two organizations that form corporations to join the consortium We're forced into joining that and made very short-term decisions on how to join that organization and the consequences of those Led to their long-term organizational failures SGCS right now has not a significant market and not a significant contribution to the X window system Even though Thomas roll was the first author of all this code that we're now shipping So what we've tried to do with the re-instantiation of the X of Xorg is put together a credible organization That will survive in the long term, and I'm hoping survive beyond even my participation in the project So when you're doing fixes short-term fixes to a short-term problem understand they have long-term consequences And the main thing is to have the main contributors to stay involved in the political process of the organization Because the political process really does Push how the technology is distributed and how the technology is implemented So if you want the price if you want the process to stay the project to stay open and the project to stay functional The technology to stay viable you need to stay involved in the political process the organization to keep the project in Keep the project shipping good free software So I think that's the end of my presentation. We had questions and comments. Yeah, a long question. You said lessons learned for the X Project, what of these lessons to also applies the Debian project in your opinion and how to fix them if so I think the key lesson that I'd like to bring to my friends in the Debian community is that Governance matters a huge amount and that if the developers abdicate their Their involvement in the political process the project will move away from their goals And that the only way to make sure the technological and social goals of the project are put forward It's to make sure the people that have those goals are involved in the process You can't abdicate responsibility and expect other developers to want to do what you want to do So yeah, keep voting keep involved Other questions We're only with seven minutes behind schedule now What do you see happening over the next few years with drivers that are currently closed sourced by provided by? NVIDIA and ATI maybe Well, I don't know. I'm working in Intel right now precisely because I'm hoping That Intel will be able to continue Shipping credible open-source drivers and as the Intel hardware increases in performance. I'm hoping a couple things as Linux becomes more viable desktop for a wider variety of people and the Linux desktop markets grows That the general interest and the general amount of money available in that market will grow as well, right? The goal that my goal here is to make it so that with having at least one open-source vendor is it We're going to be able to push the market to understanding the difference between free software drivers and closed-source drivers And have the market start choosing free software The only way that I can see to impact NVIDIA and ATI is to hurt them in the bottom line is to make it clear that their participation in the free software desktop market is Predicated on their willingness to participate in the free software environment including their drivers And so I'm hoping to pressure the market at the point where Intel's Intel's market share Which is already dominant in the desktop becomes overwhelming and ATI and NVIDIA realize that if they want to take a piece of this business They're going to have to be more open So that's my goal is to is to make that happen. I don't think that there's any other way We can force them to do it. I know the Linux kernel people are trying to shut down the ability to ship closed-source kernel drivers I think that will have I think at this point If they were if they did it completely at this point, we would end up in a situation where ATI and NVIDIA might just walk away That's my concern is that we can't be we can't push them too fast Or they'll just they'll just say if we can't play in this market. We can't play in this market We're going to give up on this, you know, tiny fraction of our overall budget overall market So I don't think we can play too much hardball right now But I think in about five or ten years when the market has grown significantly We're going to be able to play a lot nastier and push the issue a lot harder So I just want to make sure that there are some open-source drivers that it's clear that the market is interested in free software and That if they want to play they're going to play by our rules and the development of open-source alternatives to the currently closed-source drivers Do you see that as a viable option? Do is there enough information available or? can we get the information out of the cards in some way that that Those drivers will be of enough quality to compete with the closed-source drivers Actually, it is very surprising how far the R 300 project has come the R 300 project started as a Totally reverse engineering the hardware and they've done a Surprisingly good job in getting support for those cards One of the problems is that it just takes a huge amount of effort a huge amount of additional effort on top of the already Overwhelming task of putting together a credible driver to do the reverse engineering necessary to put together a real driver The other thing is that it takes a lot of time Right right now I had Intel can offer to ship drivers for new chipsets within 90 days of the chipset delivery Which means that when you buy a new laptop as long as if you buy a brand new laptop with a brand new Intel chipset You know that you'll be able to run Linux soon on it because the drivers will be available The reverse engineering process for the R 300 right the R 300 is like three years old now And it's just barely becoming usable So one of the big problems with the reverse engineering effort is while we can produce credible drivers to that hardware We can't produce them fast Right. I'm right now working on drivers for the next generation of Intel graphics You can't even get silicon for those, you know, there's no open-source driver that has one of these chips They don't exist outside of Intel right so I can actually put together drivers at ship On the day that are ready to ship on the day of ship delivery and with a closed-source driver with a reverse engineer driver It takes a lot more time So as much as it would be nice to say that we can reverse engineer and do anything The fact is that the hardware market moves so fast is that we end up with smaller and smaller Slice of the effective market as you end up with a smaller and smaller slice of the effective product lifetime Any more questions some years earlier to ship um The question is ATI used to ship open-source drivers and they don't anymore why is why the change they offer a lot of different reasons I think one of the most telling reasons is that they They were shipping open-source drivers before Nvidia was in this market before Nvidia was participating in the free software environment They were free software drivers for early 8 Nvidia chips as well It seems that at a particular point in both of these companies Hardware development cycles they somehow got it in their little brains that they had magic stuff in their hardware Whether that you know all of a sudden their hardware changed in ATI's case it was ATI purchased another company and Totally revamped their hardware architecture that are 300 series as a totally different 3d engine than the R200 series It's much more programmable. They somehow got it in their heads. This was somehow special right and so they stopped they said well There are a lot of different reasons they offer one of the reasons that we've been given Publicly, I don't know how much to believe one of the reasons we've been given to publicly is they believe they may be violating Nvidia's patents and That by shipping the driver source code would make it easier for Nvidia to stop their ship the delivery of hardware in the Windows market The other reason that we've been given is that documentation is expensive and difficult to produce That's certainly true at Intel Documentation is a lot of money to produce. I've actually been told by an Nvidia engineer that they have no documentation for their hardware Yeah that the way they do Driver and hardware development is that the hardware and software engineers work in the same area and they Collaborate together on the development of the hardware and the development of the software in tandem so that there is no actual Documentation for how the hardware works other than the driver in Nvidia's case that driver came from source code They don't have a license to distribute so in other words the only documentation. They have is the is the is the source code that they can't ship I'm just relaying question from IRC From Jonas Meyer He asks what can we do instead of using old hardware that free drivers are available for after after all we if we don't buy any Graphics cards, we're not seen in the statistics Right, so how do we how do we grow the market? We're right now the only way that you can play in the market is to buy Intel graphics hardware Yeah sucks to be me doesn't it And to buy a legacy hardware I don't know I really is the case right now the only way you can have a credible free software desktop is to buy Intel chip sets I'd like that not to be the case, but it realistically is you can okay another another Okay, go ahead. Oh, okay. Yeah Okay, I guess we're out of time today. Thanks all for coming up this morning. Amiko. Do you have a comment? Get the hang of what they they connect up. Hi Wonder if we could ship this chunk of videos to API and video marketing people saying like this guy can go to the software conference They buy Intel stuff. It's cool. It works and you can't Yeah, exactly. It is nice to be able to come up and say that we do have open-source drivers for all of our hardware so that's a nice that's a Nice thing for Intel to be able to do and I'd also like to publicly thank Richard Stallman for for for for protesting at a recent ATI event at MIT that helped me in my position in Intel significantly So thanks very much today