Linked by Thom Holwerda on Sat 23rd Dec 2006 00:48 UTC, submitted by dumbkiwi
KDE This is a response to the article yesterday on the progress of GNOME and KDE. Aaron Seigo, a lead KDE developer, sets out the current state of progress with KDE 4, pointing out that KDE 4 is on track for a release in mid 2007. "Thom points to a quote from me that our goal is to have a 4.0 ready sometime in the first half of next year. That gives us until sometime in June and I'm still thinking we can make it."
E-mail Print r 37   87 Comment(s)
Thread beginning with comment 195617
To read all comments associated with this story, please click here.
A lot of crap on that blog
by rayiner on Sat 23rd Dec 2006 06:41 UTC
rayiner
Member since:
2005-07-06

Reading the responses to that blog post was pretty entertaining. A bunch of inane uninformed rambling complaining that there is uninformed rambeling on OSNews.

While I think Thom's article was probably a bit more desperate sounding than is probably warrented, there is no doubt that there is a very big kernel of truth in his statements.

First of all, a KDE 4.0 release in "mid 2007" does not mean KDE 4.0 will be out in any practical form in 2007. The primary necessity of the 4.0 release is HIG-ifying the desktop, and that's a process that starts with a release, and continues for years afterwards (as GNOME's experience has shown).

Second of all, both Vista and OS X are in a different league technologically than any competition in Linux. No truely objective person can say otherwise. Both Vista and OS X have fully composited GUIs that are properly supported from the driver up to the desktop notification widget. Neither XGL nor AIGLX are playing the same game. XGL doesn't have the necessary underlying infrastructure, and AIGLX is a solution more notable for its convenience than its technical merit.

There is a long process remaining before the Linux desktop has something comparable to OS X Tiger from a UI standpoint, or Vista from a technical standpoint. I'll enumerate some of the main points here:

1) Proper driver-level support for a composited desktop. The DRI is not there yet. The new DRM memory manager has just landed in development versions of the Intel driver. Efficient support for hardware context switching and simultanious rendering from multiple contexts is just not there yet.

2) The window system-level support for the composited desktop needs to be improved. XRender is adequate for competing with OS X circa 2002. It cannot support the level of hardware-accelerated 2D Vista features, and that Leopard will presumably feature. Put simply: it doesn't expose the power of shaders, and that makes it a non-starter for advanced features that require that power.

3) The new capabilities offered by an accelerated, composited desktop needs to be exposed through APIs in the toolkit. Cairo is a first step in that direction, but its far from the final target.

4) All these features need to percolate through the stack and throughout the desktop. Some fancy Compiz effects are nice, but they're ad-hoc. You're not competing with OS X Leopard until the UI folks go through these new features and rationalize them with the HIG and come up with some systematic way to integrate them into the environment as a whole.

5) The stack needs to be optimized. The details of their interaction needs to be worked out. For example, when resizing a window, you want the window/composite manager to synchronize with the toolkit, and to efficiently use the memory management primitives of the GL stack. That integration and optimization isn't there yet.

Most of the pieces of what is shaping up to be a very great system are in place. However, anybody who says GNOME 2.18 + Cairo + Compiz + AIGLX is going to be in the same league as OS X Leopard is not being objective. And that's from a purely technical consideration, not some abstract "Linux is not usable as OS X" bullshit. The Linux desktop stack that will exist in early 2007 will be comparable to OS X 10.2 in technical capability. It'll be several releases and a year or two beyond that before you're looking at something with the maturity and completeness of Vista or Leopard.

NOTE: In the interest of fairness, I should point out that being late isn't necessarily a bad thing. It is very probable that Linux circa late-2008 is going to be better than OS X or Vista in the same timeframe. GNOME's UI is better today than Vista's, and is in very many respects better than OS X's (and I type these words on a MacBook). X will still be network transparent, and neither Vista nor OS X will be. X's indirect rendering model and XEvIE will allow the leveraging the user-interface potential of a composited desktop (as opposed to merely the graphical potential) in a way that OS X and Vista won't.

Edited 2006-12-23 06:54

Reply Score: 5

RE: A lot of crap on that blog
by superstoned on Sat 23rd Dec 2006 11:53 in reply to "A lot of crap on that blog"
superstoned Member since:
2005-07-07

Kwin is build to be able to use XRENDER or OpenGL, so hardware acceleration will be there for all. now many effects can't be done with XRENDER i suppose but even with linux, you can't expect Vista-like effects on Win '95-hardware...

Arthur, the Rendering engine in Qt4 is much more capable (and mature) than Cairo, so KDE doesn't have to wait for Cairo to get a decent performance. You're right in the HIG area, tough KDE 4 will have automated usability testing, and use it's framework to get more usability as well. For gnome apps, you had to spend time on problems like 'the icons in Gedit are 2 pixels further from each other than the HIG states'. you don't have such trivial problems in KDE, as the framework dictates such settings. so you can focus on the really important stuff, which hopefully will result in much faster adoption of the HIG than Gnome has been able to pull off.
Also the KDE4 HIG will be more usable for developers, easier to apply with better and more examples than most other HIG's, also speeding up adoption.

Reply Parent Score: 5

rayiner Member since:
2005-07-06

Kwin is build to be able to use XRENDER or OpenGL, so hardware acceleration will be there for all. now many effects can't be done with XRENDER i suppose but even with linux, you can't expect Vista-like effects on Win '95-hardware...

As I said, composited windowing is 2002 level technology. The technologies offered by Vista and Leopard (and even Tiger today) go way beyond that.

Arthur, the Rendering engine in Qt4 is much more capable (and mature) than Cairo, so KDE doesn't have to wait for Cairo to get a decent performance.

Arthur has the same problem Cairo does. If it goes through XRender, it can't do all of Vista's and OS X's pixel-shader based tricks. If it goes through OpenGL directly, it'll hit the DRI stack's limitations on context switching and concurrent rendering.

Reply Parent Score: 3

RE: A lot of crap on that blog
by siki_miki on Sat 23rd Dec 2006 17:53 in reply to "A lot of crap on that blog"
siki_miki Member since:
2006-01-17

Even Vista got lots of advertised GUI features axed. For example Aero Glass Diamond (vector vidgets) or most effects from that 2003 video on youtube.

Obviously their driver framework and development just wasn't ready at the time for many planned features so they postponed them for Vista SE or later OS. Avalon for example isn't very much used in Vista GUI.


True, Vista 3D driver stack is already here and probably currently better than DRI. it is focused on next-gen cards and DX10 exclusively which allowed MS developers to make advanced design quickly, but older hardware support got sacrified (talking about WGF2/DX10).

However new memory manager in DRI is proof that catch-up is ongoing (despite devs currently being focused more on low-end Intel hardware). Besides, latest generation of Nvidia and ATI hardware is supported only by binary drivers, anyway. I believe that those two should be persuaded to base their future linux drivers on DRI, even if their specific kernel- or userspace driver will remain closed-source. Unlikely to happen though.

Big problem is X protocol. As you say, xrender isn't enough. We might even need something equivalent to WPF / XAML to be able to propagate and draw purely vector widgets and application windows WITH "scene" definition (borrow few ideas from SVG, Flash etc.?). Until then, workaround will be to do demanding stuff *and* 3D-on-desktop through windowed OpenGL.

Integration of compositing effects (or 3D) is another issue as you mentioned. Currently window managers are limited as they have no useful (=DE isn't really aware of it) communication with desktop env. Also toolkits don't have slightest idea what is happening, they just issue bitmaps to compositing managers and that's it. I believe there is lots of room for improvement here. First try at cpopositing desktop integration might be seen with KDE4.

Reply Parent Score: 3

rayiner Member since:
2005-07-06

Even Vista got lots of advertised GUI features axed. For example Aero Glass Diamond (vector vidgets) or most effects from that 2003 video on youtube.

That's just a maturity issue. New effects and widgets have to go through the UI people, and Vista doesn't have the time for that. However, the underlying technology to do those things is present in Vista.

True, Vista 3D driver stack is already here and probably currently better than DRI. it is focused on next-gen cards and DX10 exclusively which allowed MS developers to make advanced design quickly, but older hardware support got sacrified (talking about WGF2/DX10).

That's a cop-out. The DRI folks couldn't have come up with a Vista-like stack even if they didn't want to keep compatibility with older hardware. Not because of any lack of technical capability, of course, but the fact that they just don't have the access to the specifications of next-gen hardware in the way MS does.

However new memory manager in DRI is proof that catch-up is ongoing (despite devs currently being focused more on low-end Intel hardware).

Barring a change of heart by NVIDIA and ATI, low-end Intel hardware is DRI's best hope. Implementing a Vista-like 3D stack on reverse-engineered drivers is an incredibly daunting task. Even given the fact that the Intel drivers solve most of the spec-access issues, hoping for a Vista-like stack by mid-2007 is silly. Thom's prediction of late 2008 is a much more reasonable one.

Reply Parent Score: 3