“Compiz is the compositing window manager that works on top of Xgl or AIGLX to enable Desktop Effects. Recently, a community developer named Quinn Storm announced that she would start a Compiz-based fork project called ‘Beryl’, citing frustration with Novell regarding getting her code fixes accepted into the Compiz upstream source tree. We called Compiz/Xgl maintainer David Reveman to get his side of the story.”
Compiz has a load of Gnome dependencies. Since I don’t use Gnome, that for me is less than ideal. Beryl does other things Compiz doesn’t.
Reveman of course pointed out that the Gnome dependencies are not hard coded. “All technical reasons (alternative configuration system, alternative decorator, xinerama…) for this fork are incorrect” etc. etc.
IMHO however pointing out the fact that people can write alternative configurators and window decorators for Compiz doesn’t necessarily help it, it out just makes it more painfully obvious that for whatever reason, they didn’t. THey wrote them for what became Beryl.
Why then? Is Beryl more open? Does it accept lower quality code? Whatever the reason, stuff I want to see gets done in Beryl and saying it could be done in Compiz too is basically useless as long as it isn’t.
The whole thing is pretty much a novelty for me, something to mess around with but not use on my main desktop, so quality of code doesn’t bother me as much as drawing in tons of dependencies I don’t want. Perhaps as it all matures I’d consider actually using it day to day, in which case Compiz might be preferable for its allegedly better quality of code. Hopefully by then a KDE windeco or non Gnome configurator will be ready.
Compiz has a load of Gnome dependencies.
I think you’re confused by the term dependency. Compiz doesn’t depend on Gnome.
Reveman of course pointed out that the Gnome dependencies are not hard coded. “All technical reasons (alternative configuration system, alternative decorator, xinerama…) for this fork are incorrect” etc. etc.
He’s completely correct. Do you take issue with his claims? If so in what way?
IMHO however pointing out the fact that people can write alternative configurators and window decorators for Compiz doesn’t necessarily help it, it out just makes it more painfully obvious that for whatever reason, they didn’t. THey wrote them for what became Beryl.
That’s not exactly what happened. The fork was based on the need (or want) for alternate configuration options. The problem lies in the fact that a fork was totally unnecessay, because these features could have been coded as plugins like everything else. So why fork?
Why then? Is Beryl more open? Does it accept lower quality code? Whatever the reason, stuff I want to see gets done in Beryl and saying it could be done in Compiz too is basically useless as long as it isn’t.
Bingo. The code wasn’t up to snuff. That is a terrible reason to fork a project, but it wouldn’t be the first time. This will all go away when Beryl collapses under its own weight of shoddy code and eventually someone will write high quality plugins for alternate platforms.
Is there anyone else out there who is totally unimpressed by Xgl, AIGLX, and Compiz?
I’ve yet to see where this 3D stuff actually helps the desktop. Transparency? A distraction. The Cube? Slower than just clicking on the workspace switcher and having the virtual desktop switch instantly. (If you play FPS’s, think “weapon change animation”.)
Wobbly windows? Don’t even get me started on that one!!!
And the way every alt, ctrl, or shift, plus function key binding gets intercepted by it. Bad, bad, bad.
And this is all diverting resources from the really important thing. And that is making the desktop easier, *less* confusing, and more usable.
This Novell Infomercial was, however, one of the few plusses:
We were just coding along, minding our own business, and then that Quinn bitch showed up and forked our project!!! (Followed by about 10 restatements of the definition of what a “fork” is.)
At the very least, this whole thing is going to have some entertainment value…
Edited 2006-10-07 18:56
I partially agree with you.
However I feel projects like Xgl can be a foundation for future real usability improvements in the desktop. Transparency, cube and the like are probably just cute tests for a new technology. Their purpose is to say “ok, here you have this possibility, why don’t you use it for something useful?”
Like with almost every new powerful tool, some real nail will be soon hammered.
And I agree with you that the underlying technology is valuable. Just not in proportion to the number of jaws that drop to the floor whenever anyone whips out… THE CUBE.
Edited 2006-10-07 19:04
I agree with you about most of the effects. I find useful the expose-like one, the new alt-tab, and that’s pretty much it. However, and I think this is constantly overlooked, the importance of XGL and AIGLX is that they provide us with a truly hardware composited desktop, something we really didn’t have before.
“””I agree with you about most of the effects.”””
Out of curiosity, I installed Xgl and Compiz on my Ubuntu system. I *think* that what I got was the Quinn version.
I thought the wobbly Windows were overdone. But then, I maximized a window… and was momentarily overcome by an atack of vertigo!!! Mon Dieu!!!
I’m a big Linux fan, but when I read about how we’re trouncing MS’s Aero with our 3D desktop, I can’t help thinking “For all that is good and decent in this world, I hope not!!!”.
Edited 2006-10-07 19:18
> a truly hardware composited desktop, something we really didn’t have before.
But what is so precious about having a ‘hardware composited desktop’?
Note that this is a very difficult thing to do right: for example having good font rendering means that the font renderer knows where the pixel are, if an application draws its text to a texture, but then after the texture is shifted or scaled by a ‘compositing engine’, then it is likely that the resulting text will look ugly..
My father is older and has trouble with his eyesight. The hardware accelerated zoom available in compiz is better than any other screen magnifier available. He loves compiz for this alone.
My mother likes the minimize animation It is more “usable” in the sense that the window shrinks to the tray. A non-computer literate user might not know that and become confused when they accidently minimize a window. (yes, this does happen).
The Mac OS X Expose clone effect is implemented better in compiz than it is in OS X. Why do I say this? One key scales all windows on the current virtual desktop, one does it for all windows for the current window (think gimp with its million windows), and one for every window on every desktop. This saves lots of time when working with many windows on many different virtual desktops.
I hate to break it to you, but this *does* make the desktop easier, *less* confusing, and more usable. Novelties such as the spinning cube and wobbly windows can be enabled or disabled. Redhat just submitted a patch for a “plane” instead of a 3D cube and it is much closer to metacity’s behavior. Also, working with a “pretty” desktop is much more fun and entertaining for new users. Just my 2 cents…
“The Mac OS X Expose clone effect is implemented better in compiz than it is in OS X. Why do I say this? One key scales all windows on the current virtual desktop, one does it for all windows for the current window (think gimp with its million windows), and one for every window on every desktop.”
[…]
Redhat just submitted a patch for a “plane” instead of a 3D cube
If by “better” you mean “the same”
Expose in the current Mac OS X release, Tiger, has 3 keys/activation corners/mouse buttons (your choice):
1. All windows
2. All windows of the current app
3. Move all windows away to see the desktop.
Multiple desktops isn’t standard in Tiger, but it is in Leopard and integrates with Expose exactly in the way you described. And, unlike the current plugin for Tiger, instead of a Cube it uses a plane which is indeed (a bit) more usable.
A little vid:
http://www.apple.com/macosx/leopard/spaces.html
(ignore the silly tagline!
The Cube? Slower than just clicking on the workspace switcher and having the virtual desktop switch instantly. (If you play FPS’s, think “weapon change animation”.)
I can only use virtual desktops when I use that spinning cube. Without them, the desktops have no physical presence, and hence I lose track of which window is on which desktop. This does not happen when they are wrapped around a cube, and I can specifically assign each side to a certain function; I then set one desktop as the ‘middle’ one (usually the one with Firefox running), and from there, I have a starting point.
Wobbly windows? Don’t even get me started on that one!!!
Have you actually used them? In a sane, conservative configuration (meaning, not too wobbly) this functionality actually serves a great purpose: it gives windows a physical quality, and hance makes them easier to manipulate with. The wobbly effect specifically responds to physical movements of the mouse, this is GREAT feedback, and GREATLY enhances the windowing paradigm. I’m not kidding.
I was like you on this one before I tried it. Then I did, and it suddenly made sense.
“””Have you actually used them?”””
Yes, I have.
I find your observations to be… weird.
But if it works for you, all I can say is “You go, Girl!”
Seriously, I found SEJeff’s post interesting. It may be the first point that anyone has made about the ability of 3D to make the desktop more usable that has caused me to stop and think “That’s quite credible”.
I strongly suspect that it is the most “boring” aspects that will end up being the most useful.
I also suspect that we’ll still be laughing at the rest of it in 10 years time.
Have a great day!
Steve
Edited 2006-10-07 19:43
“The wobbly effect specifically responds to physical movements of the mouse, this is GREAT feedback, and GREATLY enhances the windowing paradigm. I’m not kidding. ”
What, in case you didnt already know that you were moving the window? Isn’t the window actually moving across the screen a big enough clue?
People hated Windows 1.0 too. Graphics acceleration technology is here to stay so we need to figure out how to make good use of it. You may hate version 1.0 of these features but by the time version 3.0 comes around you won’t be able to live without it. I don’t see many people giving up their GUI to go back to Word Perfect and Lotus 123 on a terminal emulator.
Give it time, the killer feature for this technology probably hasn’t even been thought of yet.
Edited 2006-10-07 20:51
Well, Jon, I guess I didn’t mean to be such a 3D killjoy.
I just that I see a lot of people getting excited about things that leave me shaking my head in wonder.
You won’t find me discouraging others from pursuing the use of this technology.
I’m simply amused at the things that people are getting excited about. Bling! Bling! Bling!
Again, SEJeff’s post about the boring benefits of 3D is a quality post.
Edited 2006-10-07 21:13
The most important thing about XGL and related technologies is that we now have a platform to experiment with. Given Moore’s law sooner or later OpenGL hardware is going to be essentially free. I’m hoping that Linux moves toward this standardized API instead extending things like XRender and EXA.
Erm, we all hated Windows 3.0 too.
It’s human nature. We live in a land of excesses. The more the better. Obviously wobbly windows add nothing to the computing experience except for a “oh wow!” “That’s really cool!”, etc. from your onlooking friends. I do have to disagree with you on a few points, however. First of all, the spinning cube has definitely made my virtual desktops much more useful. Before having the ability to drag windows to different sides of the cube (a.k.a. a different workspace) I would forget I had virtual workspaces. As for transparancy, I find it handy when i’m trying to copy something from a window below the one i’m typing in. As for the concept of an “accelerated desktop” Linux has needed this for a long time. Dragging windows around on the screen is smooth as silk and without the annoying tearing that I used to see. Compiz/Beryl and XGL/AIGLX are wonderful additions to Linux and i’m looking forward to what’s coming next!!
3d acceleration eases the burden off your CPU for doing other tasks. Thats the best part…. for example when i hold a window down without acceleration, and drag it around… my cpu spikes and if i do it enough and fast enough then my system becomes unusuable. thats because gtk or metacity or xorg… is bloated… or coded ugly… but letting the graphics card help helps.
Also when you move your mouse up and all the way to the right… the windows all “tile” on the desktop and let you choose between them which is really cool usability feature… i think its called expose or something on macs. Besides… a better looking desktop is just nicer for everyday use.
I have found to many bugs to use it myself at this time though. FOr instance i cant even rename a file on the desktop by right clicking…. when xgl is enabled.
Especially if everyone and their dog wants their code included. They’ll have a place to vent.
Having a sort of conservative option is always nice.
I use XFCE4 with compiz and I at first thought is was useless until I realized you can do some cool things. Expose like applications, and transparent windows can allow you to see which window to click on under a pile of windows. It can actually improve usability of Gnome if you set it up correctly. As to the 3D Cube, well there is always the crowd that loves eye-candy. Perhaps KDE will make the 3D Cube their standard
The real usability improvement will come as those effects become more refined and better though out. For now it sometimes get in the way or can appear annoying/overused. MacOS is a good example of how to do it (Vista isn’t).
Beryl is a plugin development testbed. I doubt that they have experts to radically improve design (for now) and IMO compiz will evaluate more quickly, while Beryl will have more plugins.
What struck me is that they start complaining that things have been one sided and then proceed to do a completely one sided show themselves. Sure they were coming from the other side, and it was their intention primarily to get David’s side across, but if they’re so concerned about being even handed you’d think they’d at least ask someone from the Beryl side to comment before sending it out. Also when a show related to Novell interviews someone who works for Novell they might want to preface it with a declaration that this is probably not unbiased reporting.
Aside from that the “why can’t we just have one really good compiz” thing sort of annoyed me. One thing I like about Linux is choice. I can choose Gnome or KDE, Firefox or Konqueror (I chose KDE on the last question), and now if I want Beryl or Compiz. Having more than one program do the same sort of thing allows people to choose the one they like. Having only one “really great” option allows everyone to hate the same thing, in equal measure, but for different reasons.
I am all for choice too. However, my concern with Beryl is that its popularity due to its features will cause it to become a “standard.” How many pieces of the linux system do we routinely shake are heads at and wish they were designed better from the start? How many limitations will we regret having not done right the first time? Personally, I say stick with the quality code and build it *right* rather than now.
Howdy
I really think you`ll find that big distros will stick to the “official” tree backed by a large corperation and that individuals will probably use the “storm” version.
As for forking = bad well just think about those developers stopping all development due to their frustrations, basically everyone looses in that case and some duplicated effort is not such a bad thing really and if you look at ALL the Linux distro’s kernels for instance you`ll see that sometimes 2 or 3 ways of doing things gives another an idea that eventually emerges as the “right way”.
“””What struck me is that they start complaining that things have been one sided and then proceed to do a completely one sided show themselves.”””
Plus that fact that it was just so… corporate. They sounded like they were struggling to sound so… hip.
I’m sure that every word was reviewed and approved by the legal department.
David doesn’t have to defend himself or Novell.
Neither does QuinnStorm.
We’re talking about GPL code. If one group of programmers isn’t satisfied with the existing project, they have the right to modify the source and redistribute it as they see fit. As long as they pass their modifications back to the community.
I have tons of respect for Reveman, but I don’t see the downside to having Beryl “compete” with Compiz.
While this project isn’t as big of a deal there are big disadvantages to forks in many cases. As much as we all love the choices that Linux provides it also is one of the biggest problems for it (and free software in general). Outside of some kernel api’s a software developer is left with trudging through countless API’s and platforms all designed to do the same thing.
In this case instead of the contributing talents unifying behind one 3d framework it’s being split into two, evermore incompatible frameworks. I know a lot of people think this is what makes linux great but it seems to often lead to slower progress, redundant effort, a fractured community, and more confusion.
If your freedom to choose means freedom to pick the only option, then it’s worth nothing.
I think the “useful” effects like zoom, miniaturizing and exposé-clone are quite nice, but as long as they require a special window manager or hefty extension to a window manager these projects are on the wrong path I think. My wish is to have it all in X11, defining the settings xorg.conf independent from the WM and then just have any key binding I like. Just imagine Fluxbox with a useful task switcher and exposé…
This show reminded me of Wayne’s World or something… two idiots blathering with some heavy metal in the background. What’s worse was probably the sound quality of the phone interview, but I guess David R was saying something like “I don’t see the need for this fork, but I respect it.”
.. which of these get BumpTop work well with it first. Its one thing to be stable/clean and another to have functional usefulness as well as facilitating inovation. If you take too high of a stance on selective inclusion progress slows down. Look at Xorg vs XFree86 or Debian vs. Ubuntu or KDE vs. GNOME. What I think would have been best if they worked together to develop and establish a framework/platform in common that they could all work on improving and extending. Forks I think only lead to duplicated effort and dualing camps.