On Fri, 28 Feb 2003, Jon Smirl wrote: >> I don't see 100 unpaid hackers hacking feverishly on >> anything X > >Obviously you wouldn't see 100 people working full >time
Obviously. Not even part time. I doubt you'd see more than 20-30 people "working" on it at all. And by that I mean making any significant contributions, not a oneshot bug fix. I mirror what Gareth Hughes said. There aren't hundreds of performance obsessed hackers itching to hack on DRI source code, wether it is the existing open source code, or wether the sources for both ATI, and Nvidia's drivers were released. In fact, I'll go one step further, and say that I believe if ATI and Nvidia *both* released the _entire_ source code to their drivers, and published their complete video hardware specifications on the front page of slashdot, that practically zero work would be done with any of it except perhaps by existing people already members of the DRI project, or hacking on DRI a fair amount on this list already. I bet it would draw next to zero new developers. I also doubt that any performance improvements would be made to those currently closed source drivers any time soon. More likely, the code would be used to improve all video drivers, including the Matrox ones, and other video hardware as well. That would take a LONG time to do too. It would probably take enough time for experienced DRI people alone to even get a hold on the full driver source code base enough to make use of it. >but you might get detailed bug reports or patches from 100 >people. I know I get patches from all over the place for code I >have written. Detailed bug reports? LOL. 95% of all XFree86 related bug reports from people are pure useless garbage. Missing tonnes of information, and quite often you have to dig the information out of the person, and they get increasingly frustrated with you and upset and flame you so you don't even want to look at their problem anymore. Patches? Not many. A few patches come in (specifically related to DRI that is), yes, but they are very few (IMHO) compared to the work the DRI and XFree86 projects do. Quite frankly, the collective code of the kernel interfaces, DRI, Mesa, etc. used in implementing open source 3D is quite complicated in nature, and there aren't 100's of people out there IMHO who grok it enough to be able to make as many contributions as you might think. Having more code out there doesn't mean there are going to be more people hacking on it. >A couple of months ago I wanted to make a few changes to the ATI >Rage128 framebuffer driver. It took me a month to get ATI to >give me the register specs. I still can't get the 3D spec and >this is for five year old hardware. I fixed a couple of bugs in >the framebuffer. I just left the others alone since I can't get >them to tell me how to reset the card from protected mode. You might not have been able to get them, but many of us do have them. While it would be nice if hardware vendors would release specs openly, it is their decision. The fact that they allow ANYONE to have them is FANTASTIC, or we would have little to no open source drivers at all. Be thankful to those companies that do allow some developers to have access to documentation. >I just don't understand what is to be gained by keeping the >Rage128 hardware programming spec secret. After all a device >driver for a board has the best copy protection in the world. That's their business really. I've got the Rage 128 specs, and many other X developers do too. If other people can demonstrate their abilities to work on X or DRI, and show enough initiative to work on the code *without* the specs, they are much more likely to get them. I know several people who hack on the ATI drivers who do NOT have the specs, and they manage to do pretty good without them all things considered. ATI would probably give them the specs based on their track record if they asked. >We don't know if NVidia or ATI have incorporated 3rd >party code into their drviers. I am pretty sure that it has been acknowledged that they have. >There are other solutions. >1) ATI could simply open source their hardware spec >and let us write the drivers To be honest.. ATI provided many developers with Radeon R200 specifications prior to the hardware even being released. I don't know specifically how many people had them or have them now, but I do know that most if not all of the developers who have them, have full time jobs, and do not have time to spend 8 hours a day working on video drivers. The largest development efforts will come from people who can afford to spend hours every day working on them, and that is only most likely going to happen if it is funded development work under contract, such as the work that Tunsten Graphics is doing for The Weather Channel right now. Or the Intel i8x0 driver work that's been done. There is open source work being done by volunteers, and I think it is fantastic too. The Mach64 project, talk of Savage driver, and others on the way for older hardware. I just don't think it is realistic to expect that source code is going to materialize merely because specifications are made available. The 3Dfx Voodoo 3 and Banshee specs are available, as are docs for other 3D hardware. Who is working on that right now? 3Dfx released the source code of Glide3 for example. I dont think a single line of code has been written for Glide3 for 2 years now. Don't get me wrong here, I want DRI to succeed, and to be as open source as possible. But I think people need to be realistic about it and get off the fantasy clouds they're living in. My general experience of seeing people ask for specifications for various hardware, more times than not the person does nothing with them, or they fix one bug that bothered them then go away. That's either with specs that are open already, or with specs that they were able to convince a vendor to give to them. It'd be great if specs were all opened, but I don't think that there will be 100's of developers standing in line to hack on code if they were published on tomorrow's slashdot front page. There would be a very small handful. And hopefully they'd put together a successful project around it, but it wouldn't happen very quickly IMHO. Not without funding. >2) ATI could shift resources and contribute to the DRI >code base instead of working on their own. Let me play devils advocate... What would they get out of it? >3) ATI can license their patents for royalty free use >when developing drivers for their hardware. Other uses >would require fees, like the new W3C patent position. I'm not talking about ATI's patents. I'm talking about other company's patents to which ATI may have licensed perhaps to use in their own hardware or software. They may not have legal rights to divulge this information. I fully believe that vendors can benefit the most from having open source drivers and specifications. I do not however believe that if a vendor releases specs or source that it will result in an overnight sensation of a stampede of new developers entering the foray. I also understand the various complex webs of legal issues such vendors may face due to licensing of patents, cross licensing agreements and other legal issues, and I respect their positions even though it would please me greatly more than anything to see all specs and sources opened up. Someone should really turn this topic into an FAQ we can point people to every 3-4 weeks when the topic comes up again. -- Mike A. Harris ftp://people.redhat.com/mharris OS Systems Engineer - XFree86 maintainer - Red Hat ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ Dri-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dri-devel
