On Wed, 22 Oct 2003, Alex Deucher wrote: >> Oh, wish I could. It's preety annoying to see that your computer gets >> faster with a oldie voodoo 4 pci than with its own agp 8x savage, >> just because of the driver... > >One of the things that always gets me about computer journalists and >reviewers and even users to a certain degree is their tendency to blame >developers. Absolutely, it sickens me. Thankless bastards. >If you read a story about linux in PC Magazine or on ZDnet, they >always say something like "the opensource driver developers need >to get their act together and support more hardware, or support >hardware better or linux will never catch on." I hate to send them a gigantic wakeup call, but Linux *has* caught on, and over the last 10 years, each year it has attracted more and more people. The current trend is that that continues to increase at a higher and higher rate. As such there is absolutely zero statistics whatsoever that Linux "is not catching on" or that Linux's lack of support of a given piece of hardware is preventing the OS from growing new users and customers. Those "linux will never catch on" statements are pure bullshit FUD, with no basis of fact whatsoever. But then, I'm preaching to the choir now anyway... >No one ever wants to blame hardware vendors. Indeed, and it is the hardware vendor who makes the hardware who is ultimately responsible for providing support for their hardware if they want to sell it to Linux users. One could say "Hardware vendors need to get their act together and support their hardware in Linux better if they want their hardware to catch on". That is a much more realistic statement than the one often used which you quote above. >The fact is, there aren't that many developers working on >opensource drivers. Most do it in their spare time. Hardware >venders employ people full time to develop drivers for windows. Exactly. Open source drivers are mostly developed by volunteers hacking on the code for free in their spare time. Some developers working on drivers are employed by a Linux distribution vendor or some other company, but the number of developers that is true for is far far less than in other operating systems, and the responsibility lay on the hardware company themselves, not the OS vendor, although OS vendors do spend a lot of resources on supporting hardware nonetheless, and often with zero help from the hardware vendor. >No wonder the hardware is well supported and generally bug free. >Beyond even developer time, that fact is, many hardware venders >don't release databooks for their products. For video - barely any vendors release databooks for their hardware, and even then they are not 100% complete usually, and are generally available under NDA only (which is much better than not at all). >without them developing a driver is either impossible or 5x >harder than it would be with databooks because the developers >have to figure out how the hardware works from trial and error, >then develop a driver on top of that. For video hardware, without any help from the vendor and without any hardware specifications from the vendor, hardware support is non-existant period. The best a user of such hardware will get, is using the "vesa" driver which uses the video BIOS, and that is very slow unaccelerated video which is very buggy in general, and very limited. >It is also oftentimes the developer that also supports the >drivers. Hardware venders have whole departments that deal with >product support. Exactly, and volunteer developers have real jobs, real families, and other priorities. A volunteer developer volunteers their time when it is enjoyable for them, or for some other motivation. End users demanding support and ranting and raving, or bitching about something that isn't supported, or bitching about bugs in a given driver generally do more to make volunteer driver developers go play a game of football or go to the movies on Friday night than give the slightest flying f**k about the user's problem who is bitching and complaining. For example, if someone wants me to hack on something in my personal spare time, and I'm not interested in hacking on it at all, or I don't have the time - they have no right to bitch at me. They can however come to my house and wash my dishes, vacuum my living room, and cook supper while I debug their problem on the card they brought over with them, for their night of being mharris's maid. <evil laugh> >So, if you want good support in linux, stick with hardware venders that >help out opensource developers. Developers are more inclined to work >on projects with receptive venders. That's why ati hardware is so >well supported. Exactly. ATI sends me so many patches that it is a pleasure to work with them on every level. In fact, they've sent so many patches lately that I haven't yet had time to even look at them all. ;o) Mind you, most of these patches have now been included in XFree86 CVS, which is more important for the community at large. I hope to get them integrated into 4.3.0 as well for future erratum once i have time to investigate things in detail. VIA also has been very good to the community lately, and are now maintaining their own driver, and providing code, etc. Most other video hardware vendors don't participate in the community very much at all nowadays. Some silently contribute code once in a while, or fund development of their drivers, but those are the exceptions to the rule. >If you want to use S3/VIA's savage driver, you'll have to use xfree86 >4.2 for the time being. Unfortuntely developers just don't have enough >spare time or lack the hardware to port the driver in a short amount of >time. Indeed. People don't realize just how much work is required to do that, and that all of this work is UNPAID FOR, and done BY VOLUNTEERS in their SPARE TIME, of which they might have 500 other things they do in their spare time as well. Support might take 3 months, 6 months, 12 months, or might never exist in the long run. The Mach64 3D support for example was begun well over a year and a half ago by numerous developers volunteering to get it working. That work was not yet completed, wont be included in XFree86 4.4.0 and the project seems to have slowly come to a halt, which means Mach64 might never have supported DRI 3D acceleration support. >> Anyway. I took a look at Tim's page, looking for updates on its >> driver, >> or even hoping that he took the via driver and fixed up himself :-) >> that's not the case, but it indicates a driver for redhat 7.3 and 8. >> Nothing for redhat 9 yet, in fact, it doesn't even compile. >> > >The "via" driver and the savage driver are for different hardware. the >xfree86 "via" driver is for the CLE266 integrated video hardware from >VIA. the code for that driver was also released by via for xfree86 >4.2. Alan Cox and several others cleaned up the 2D driver for >inclusion in xfree86 cvs. Alan also started porting the via 3D code to >mesa 5.x, but never got it finished. it now resides on a branch in >cvs. Alan's quite busy these days, but I believe he's still working on that. One can imagine however that Alan has a lot of irons in the fire almost always, and so he obviously can't spend 8 hours a day working on Via 3D support. ;o) >The savage driver that S3/VIA released is also based on Tim's code but >also needs to be ported to mesa 5.x. it too is in a branch in DRI CVS. Yep, and it'll probably all be ready for usage for XFree86 4.5.0 or whatever is after 4.4.0. People more or less have to either: 1) Patiently wait 2) Purchase hardware that actually has 3D support right now 3) Volunteer to hack on support and actually *DO* it, not just talk about it or rant and rave to other volunteers 4) Get 1000 other people who are also interested enough in the hardware being supported to each donate $100 USD to a fund, and pay a developer to do the work. 5) Email/telephone their video hardware vendor and request official Linux support politely. Get 5000 other people to do so as well. 6) Vacuum my rugs and do my dishes, cook me food, massage my feet, run errands for me, etc. (No, I don't want money, I want clean dishes! Make 'em shine! ) ;o) TTYL -- Mike A. Harris _______________________________________________ xfree86-list mailing list xfree86-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/xfree86-list IRC: #xfree86 on irc.redhat.com
