> Am 14.12.2010 21:08, schrieb Adrien Destugues: > >> I suspect a large percentage of the poll-takers don't even know if > >> their hardware is supported by one of Haiku's existing video > > > drivers, > >> and as a result were unable to make an informed choice on that > >> particular question. > > > > I think it's cheating to ignore the result telling people 'anyway, > > you > > didn't understand the question'. A lot of people have or had > > problems > > with video mode setting. I know my work on intel_extreme helped > > some of > > them (including myself). > > > > People were allowed to tell they didn't understand the question. > > So, > > why would they have checked an answer anyway ? > > In my original mail I had not claimed people didn't understand the > question. I did raise some issues, though, that might exist with this > particular poll result, but I don't feel like repeating them again. > I read it, and the other points are accepted :). > > For intel_extreme and radeon the problem is not so hard to solve : > > the > > hardware is well documented, and there's xorg code at hand for > > reference. For nvidia, I don't think there are available specs > > (unless > > I missed something), so it's a bit more difficult. > > Yes, it's more doable with specs. But you neglect the issue that > someone > actually needs to do the work. Whether it's possible at all is less > interesting. The whole context of the poll is how long should R1 be > delayed, there is no doubt that it would be nice if Haiku ran on all > hardware with zero issues. I can work on it (after all, its more related to my school courses than i18n). I already did write drivers for the hardware I have access to. This is the locking problem here, more than on other issues : we need a dev with both knowledge and hardware. Personally, I would rank video drivers higher than ARM port on my wishlist ; so it would make sense that the NPO funds buying video cards. Some of the drivers are simpler to get done than others. For example the intel HD one is only different in the way of setting up the screen. The driver itself wouldn't even need a single change, only the accelerant would be affected. We already have a small part of a Radeon HD driver started by Clemens at BeGeistert 022, it could be completed to read the EDID info and get awhole other family of cards working. This gives an upgrade path to everyone. Remember the problems of TuneTracker systems finding new hardware compatible with BeOS. If we allow at least one working solution, they can switch to Haiku, and other 3rd party devs (commercial or not) will also join in. They will then help filling in the other missing parts. > It is interesting you mention your T60. Have you tried removing the > intel driver? Was the native panel resolution not in the VESA list? > At > least I remember a lot of tickets mentioning that running in VESA > works > just fine (via save mode options). Sure, there are some people who > then > can't pick the native resolution and it's a no-go on LCD. My whole > point > was that the poll result itself does not tell how many of the people > who > voted pro native drivers would have been helped exactly by removing > the > native driver and running in VESA, for example. You simply can't tell > by > the poll results, but this exact sitation is definitely real. The dell inspiron worked fine in VESA mode. It had the right resolution, but was flickering at 30Hz, with the intel driver. The T60 was at 1280x1024 instead of the expected 1440x1050, with both drivers. VESA is a way to solve part of the problem, but not all of it. And the fact that you get variable results even on the same graphics chipset; or even the same computer model, means when you buy a computer, the only way to know if it will work is to try it. Being able to tell we support a list of chipsets (for example, intel Hd and Radeon HD), mean people can buy avideo card and know it will just work. Right now, we only have drivers for cards that aren't shipping anymore, or will soon disappear. Anyway ; on the other hand, if we can always boot in VESA, it gives people a way to go on the internet and get an updated driver that would be released after R1. So the availability of VESA lowers the pressure on the other drivers, even without replacing them. > > > With the specs available, and the hardware at hand, writing a > > driver is > > not a very long task if you only do mode setting and framebuffer > > access. I think we could get the hardware founded by the NPO or > > even > > lended by users willing to get the driver done. In some cases it > > may > > even be possible to work remotely using ssh to a user's box. > > You are not being realistic about how hard it is to find someone to > do > this work in the first place. If it was so easy, don't you think we > would have more native drivers already? As I said, it's more of a matter of getting the hardware. Needing more manpower is a valid point for any of the proposals in the poll, so we may as well drop them all out and release R1 tomorrow, in this case.