Originally Posted by russofris
I think you are mostly right. What I hope is that Wayland will make work easier for all the developers. Certainly a big win is that it brings a large part of the linux graphic stack up to modern standards (and should make Peter Hutterer happier), and make things easier to work with in the future. So, it is possible that proprietary drivers get ever so slightly more streamlined.
Can't nvidia just help improve nouveau and mesa? I mean, my understanding is that the nvidia blob is very tied to the X11 architecture and it does bypass a lot of things that the kernel provides (DRM, etc).
Originally Posted by liam
So wouldn't it be easier for them to just work on nouveau instead of having to port the whole blob to another windowing system? Is that even possible with Wayland?
nouveau already seems to provide a good foundation, given that it's tightly coupled with the kernel and other components (KMS, DRM, DRI2, Gallium3D, Mesa), it seems like it just needs more improvements with performance and also mesa needs to catch up with things like the latest OpenGL.
So wouldn't it be more reasonable for nvidia to just invest in nouveau?
Last edited by asdx; 05-09-2012 at 08:35 PM.
Nvidia officially supporting Nouveau would be tantamount to them officially endorsing a reverse engineering effort of their products (namely, their binary drivers and their hardware).
Originally Posted by asdx
I'm not sure they are particularly motivated to try and take legal action against the Nouveau folks as-is (in fact I rather doubt they would), but neither are they going to come out and say, "Yep, go ahead, RE our drivers; that's a good way to develop open drivers for our hardware! Release some open documentation while you're at it!"
That kind of attitude is just NOT the way Nvidia works as a company. They don't "get" open source at all, save for maybe half a dozen people in the entire company. And those people aren't the senior execs calling the shots; they're people "like bridgman but at Nvidia" (middle managers or worker bees). They can't turn the gears of the company; they are the gears.
It's too bad, because both of Nvidia's major x86 graphics competitors (Intel and AMD) are a LOT more open about publicly supporting official open source graphics drivers. But Nvidia would never do that, whether the implementation of the drivers were "clean room" or "reverse engineered" or even "officially supported by Nvidia". They just don't like open source at all, because there's a risk of their precious "IP" leaking to AMD and giving AMD an edge. Ohnoes!
(Aside: The final jab in the stomach for Nvidia is that, despite remaining as selfishly proprietary as possible, the community's RE efforts have done a great job of providing AMD easily-obtaining information about how their competitor's hardware works, and they could probably use this to help them improve their own products if they wanted to. Might be a contributing factor to why AMD's harwdare is so outstanding and extremely competitive / better than Nvidia's since about the HD4000 generation!)
Pray he doesn't modify the deal anymore
This reminds me of the scene from Empire Strikes Back where Lando is talking to Darth in reference to changes to the supposed deal they made. I used to think only Fedora was a joke-in-progress as now it seems Ubuntu has become.
So I don't believe users can keep up with Canonical's ambitious ideas.
It's generally the amateurs that come to the table with the bright idea of throwing out everything and starting over only because they are too lazy to comprehend the code-base. You've seen these individuals at your workplace. So eager are they to suggest a new tool to redirect your attention from their lack of progress.
I forecast RedHat staying with Gnome 2 for sometime. That means CENTOS and Scientific can still reliably provide for my own needs. Look how it is becoming.
"Canonical is just a false prophet offering the Olive Branch dipped in poison.", me
If this company seriously cared about humanity we'd still be using gnome 2.28 with simple bug-fixes and back-ports.
Too bad I don't have a fortune to support my vision of a great product.
It's not "starting over", it's "evolution" -- they are different. Starting over means wiping the table clean and re-designing everything from scratch. Evolution means taking what you have and reusing as much of it as possible while creating something new on top. Gnome (and to a lesser degree, Unity) have performed "evolution", not "starting over". Same with Cinnamon (which you may like as a user of Gnome2, BTW).
Originally Posted by squirrl
First of all, GTK3 and its underlying GLib library are pure evolutionary compared to their predecessors. Some 98% of GTK2 code can simply be recompiled almost as-is for GTK3. All the old GLib middleware functions are there, and in fact, GLib never even had a "3.0" release -- its version number is still in the 2.x range, so they haven't even broken ABI.
Gnome 3's features are in direct response to repeated user requests for a desktop that can takes advantage of today's 3d graphics cards (the more long-standing request), and also which can work well with a touch interface (the much more recent request, which can be seen by e.g. users of Ultrabooks, which are x86 thin laptops with a keyboard and a multitouch screen).
I'm sure that if you were using an Ultrabook, you would agree that Gnome 3 is more usable than Gnome 2. So what actually happened is that Gnome changed its target market away from traditional desktop users, towards mobile computing and touch interfaces and people who like eye candy.
This is, naturally, going to alienate people who prefer to interact with their desktop primarily using a keyboard and mouse, and who don't give a crap about eye candy. But the perceived demand for eye candy and touch interfaces is so overwhelming that, essentially, Gnome doesn't care that people like you and I are still wanting the old interface.
Well I'm sure some of the contributors to Gnome care. But overall, sacrifices in the design have been made, in order to please touch users / small screen users / mobile users, to the almost universal displeasure of workstation users. That's just the way it is. It's no big conspiracy; it's just that the projects naturally want to support the user base that is going to have the most users... and they think that is or soon will be people on mobile devices / tablets / ultrabooks with touch interfaces.
BTW, Red Hat is one of the main contributors to Gnome. You can bet your arse that RHEL 7, which will be based on something like Fedora 16 or Fedora 17, is going to ship Gnome 3 (because that version of Fedora does). Tough luck. You'll just have to keep using RHEL 6 forever I guess, long after its 7 years of security updates have expired...
I'm evaluating 12.04 in a VM to see if I want to upgrade and so far I'm not sold on Unity. I think I might be able to make it workable in a multi-tasking environment if I can get used to it. I don't know.
But what amazes me is just how non-configurable the thing is out of the box. I guess I'll have to download all kinds of tools and crap just to get some kind of sane setup going. Kinda lame.
I can't speak for Gnome shell -- since the few times I've tried it I've found it to be 100% unusable for anything more than just staring at the wallpaper -- but Unity looks almost embarrassingly identical to OS X. Between the global menu, big-honkin' dock and the way System Settings is laid out. It's one thing to be inspired by another... but making a carbon-copy just makes one look bad. Not that OS X doesn't look nice and polished... but I still find it kind of hard to use for anything more than basic work.
Dash is kinda cool though... at least the search mechanism. Some of the other stuff like automated backup, ubuntu one and the software center have improved nicely over 10.10 and 11.04.
Gnome2 is getting long in the tooth and I do want something modern, fresh and slick... even if it's different. So I might end up biting the bullet.
Can someone explain what this will mean for the future of Unity-2D in 12.04? If 12.04 is an LTS release, would they drop features during the support period that were present in the initial release?
I would say it's actually the opposite of what you write here. That because nvidia bypasses parts of the usual infrastructure, they are *not* tied to a specific architecture. They have their own thing, and they can adapt it to whatever. The whole driver is an adaptation from Windows to Linux, for example. But for now, there's just no incentive to adapt to anything other than X. Why should they, the Wayland protocol isn't even finished yet, and what's already there is subject to change until 1.0 is released, and even then there will probably still be changes.
Originally Posted by asdx
Wait, wait, are you saying here that Wayland devs are pandering to Nvidia, and that if they didn't, they'd be a lot further along? Huh?? You are aware that Weston, the reference compositor, only works on KMS/DRI2, right? At least I think that's still the case.
Originally Posted by asdx
So what makes you think they *aren't* doing what you say here, what makes you think they aren't making Wayland the best it can be? The Wayland protocol itself doesn't care how the mode is set, just that it is set, so implementations can very well work on something other than KMS. But that's not pandering to Nvidia, that's just plain good sense. If you think Nvidia is holding back development of Wayland, that's just dumb.
Basically, wake up and realize that there is a real world out there, outside of your idealistic and naive views, and that this real world has real needs that have to be met, no matter how much you're saying "just get rid of X already". Like I said, why haven't you done so yet on your machine?
I hope they fix the problem with Unity 3d windows not redrawing over VNC first.