If this is your first visit, be sure to
check out the FAQ by clicking the
link above. You may have to register
before you can post: click the register link above to proceed. To start viewing messages,
select the forum that you want to visit from the selection below.
I didn't mean they were hostile or anything. Just that I'm not aware of any efforts to accommodate blobs yet and if Wayland doesn't show any interest past 1.0, this could become a problem. But as long as Wayland and others agree to tackle blob support together, we'll all be in heaven.
PS I doubt nvidia's blob was written for X.org only because it seems to be working fine on Windows, too.
The Wayland devs are ready and open, it's Nvidia who has to step in and start cooperating, the source code is there, just read it and propose better solutions if you (Nvidia) have ones. But the cooperation must be reasonable, e.g. Nvidia can't ask to remove KMS and/or EGL (or any other core concept) in favor of something that suits Nvidia better.
Nvidia's Unix/Linux blob was written only for X.org, but of course the driver core is platform neutral. Though non-core stuff is X.org specific - it has XRandR, COMPOSITE, xlib etc etc support which is not needed under Wayland either at all or replaced with the corresponding (and simpler) Wayland solution.
I'm sure creating a driver for Wayland is way easier than for X.org because by today's standards the latter is an old, sophisticated and wacky standard.
Wayland doesn't have a good or bad attitude towards blobs, it's about blobs not working under Wayland because they've been written to work on Unix-like systems only under X.org, hence to make the blobs work with Wayland the best way would be to implement EGL & KMS support, that is, the stuff that Wayland relies on. There's no bias nor artificial barriers.
Can binary drivers implement KMS without using EXPORT_SYMBOL_GPL?
Binary blob drivers do all their memory management and modesetting in kernel space. They already have "KMS", it's just not the open source KMS which is used by Mesa drivers, but their own closed-source re-implementation.
So nVidia driver and fglrx already have KMS and fglrx even already have EGL (but this implementation probably require testing with Weston (or other Wayland compositor) before it will be available in regular fglrx releases).
please stop providing us with your blob. Instead, please provide specifications to improve the nouveau driver.
Please, try to think before you speak. ATI/AMD has been providing specs for years and the open driver still sucks. Do you really want nvidia's blob to go away tomorrow?
Try to put yourself in their shoes: they have a binary core that can be made to work on Linux, Windows, Solaris and BSD. Why treat Linux differently? Does that make business sense?
How does the radeon driver sucks exactly? At least they don't break on suspend/resume or in every new Xorg/kernel release like they do break on the blob. Radeon is also open source so developers can add things like Wayland support very quickly; in fact, Radeon and the other DRI/KMS drivers already supports Wayland since day 1.
Binary drivers will always be behind because with binary drivers then users have to wait for corporations (that don't care about Linux) to add support to everything the community comes up with, and that obviously won't always work, since it's not in the best interest of a corporation to support the Linux community.
I think you got a little confused there. ATI went open source because their Linux drivers sucked for years. Compared to that, the open driver may be a bit better, but I doubt it supports OpenGL4 or power management - after what? 3 years in development?
Nvidia on the other hand has a pretty good blob. If they released specs together with ATI and nouveau would on the same level with the AMD driver, nvidia's blob would still run circles around it. Sure it doesn't do KMS, but is that preventing you from using the video card? It doesn't do Optimus either, that could be a bummer for some. Still it's one useful feature missing.
Originally posted by asdx
I'm not going to put myself in the shoes of a corporation that doesn't care about Linux.
Then why do you keep asking them to listen to you?
If they don't do what you think they should, it doesn't mean they don't care about Linux. They wouldn't have the best blob for Linux or have the professional market cornered if they didn't care.
Let's see when Wayland is ready, because right now it's mostly unusable and lacking crucial features even on Intel drivers.
Could you please state an example of how Wayland, the protocol which specifies how a client should talk to a compositor and vice versa,
is "lacking crucial features"?
Are you possibly talking about the lack of relative pointer events?