A Look at the Compiz and Beryl Merger

invisibastard writes to mention that Linux Tech Daily has an editorial on the merger between Compiz and Beryl. "This state of affairs was a shame. Something that was finally getting the general public excited about Linux, the 3D desktop, was wasting time with duplication of effort and fighting. There were concerns about the long term viability of Beryl. The perception in the community overall was, Compiz = old and stale, Beryl = fresh and exciting. This despite the feeling in the Compiz community that the "real work" was being done by David Reveman and Compiz, and there were exciting things with Compiz core (like input redirection, etc...) on the horizon."

Comments

slashdotted

[Harik]

corel cache is up here.

Here's TFA

[rrohbeck]

I got through after a number of retries...

Editorial: Compiz and Beryl Merger

It isn't official yet, but Compiz and Beryl are merging. For the last few weeks I have been following the mailing list discussions on this topic. A lot of the work has been started. It is sort of unofficially announced, so I feel now is as good a time as any to comment. First some back story:

The war between Compiz and Beryl has been entertaining if counterproductive. Originally I planned to interview Quinn (Beryl's unofficial leader) about the Beryl project. That turned into an interview with the team that never really got anywhere. I dropped the ball. My feelings at the time were typical of those in the community. Beryl seemed to be this fantastic project that saved Compiz from being boring and a slave to Novell. They launched a beautiful website. It was exciting to see the frequency of their releases. At the time, I decided to check out Compiz to see what it was up to. It was surprising. Their forums were very helpful and positive. The more I read, the more I realized that I had made a mistake. There was more to the story than I was aware.

The communities were getting along a lot worse than I had realized. People in the Beryl camp dismissed David Reveman (creator of Compiz and XGL among other things) as a bad coder. Compiz dismissed Beryl as hacky code. Personal attacks flew around. Through decisions made with (hopefully) good intentions, like the insistence that Beryl code be GPL (thus unable to move upstream to the MIT licensed Compiz core) or the desire on some Beryl developers part to rip apart the Compiz core and " improve" it, it looked as if the teams were hopelessly split.

Meanwhile, Beryl continued to grow. Resentment grew in the Compiz community. One estimate was that Beryl used 95% Compiz code while taking all the credit. YouTube filled up with tons of spinning transparent cubes and burning windows. Any Digg story mentioning Beryl received a lot of Diggs. Flamewars in comment sections broke out regularly. Things reached a low point when a frustrated Compiz community member hacked the Beryl site.

This state of affairs was a shame. Something that was finally getting the general public excited about Linux, the 3D desktop, was wasting time with duplication of effort and fighting. There were concerns about the long term viability of Beryl. The perception in the community overall was, Compiz = old and stale, Beryl = fresh and exciting. This despite the feeling in the Compiz community that the "real work" was being done by David Reveman and Compiz, and there were exciting things with Compiz core (like input redirection, etc...) on the horizon.

It was a pleasant surprise to see talks of a merge start to show up on the mailing lists. This article by Kristian Hogsberg seemed to kick it off. The talks so far have been bumpy. There are fights about whether to rename the communities. There are heated discussions about what the merger means and where things should go from here. Old wounds have been reopened. There are complaints about the egos of the developers in the forums. At one point, reading a twenty-four page forum discussion, I wondered if the merge was a good idea after all. Little by little things seem to be working out, though. Quinn mentioned in one forum post that the fork was a mistake and regrettable. It takes a big person to make an admission like that.

I have to hand it to both communities. This is a brave and bold step. Not many of us can check our egos, put hurt feelings aside and move forward. The road ahead won't be easy, but the benefit to the Linux community will be immense. Energy won't be wasted on fights and duplication of effort. Confusion over what to use will be eliminated. Hopefully more effort can be spent by the distributions on getting the combined product packaged properly (How many times can I install a distro and the 3d desktop only to have no window borders in KDE?). The discussions I read are passionate. It looks like the project will be a meritocracy,

Re:Leopard

[spun]

Springy windows are a neat effect, and tunable so they aren't distracting. You can tune the friction and spring strength so they barely wiggle at all. But they are also incompatible with the window-snap module, which I prefer. A separate but similar effect is the focus-shiver effect, which I find very useful, as it makes the window that receives focus shiver a little, calling attention to it. The really useful window-movement plug-in is transparency, to make windows semi-transparent while dragging.

For a really fun time, try turning on springy windows, turning the spring strength all the way down and the friction all the way up. Then try to drag the window. You can stretch it practically all the way around your desktop cube.

All in all, this reminds me of way back in the day when Enlightenment (the window manager, kdawson, not the metaphysical oneness-with-all thing) first came out. Everyone started making these obscenely complex themes showing off how cool E was. Then it seems like everyone uttered a collective "Meh," and went back to FVWM. I did, anyway.

Beryl/Compiz does have other modules that enhance functionality such as tiling/cascading, and some that are mostly for show but have some use, like trailfocus. Perhaps the most interesting thing is that all the effects are scriptable, so that different effects or placement schemes can be applied to different classes of window

Re:Good for them

[grammar fascist]

I don't really know that it's a weakness. Having pretty candy might attract some new people to linux, but I for one have never been found thinking "gee, if only I had some shiny translucent three-dimensional, shadowed, foldable, fancy interface".

I like my bling, I run Beryl, and I've got windows that go up in blue flames when I close them. (Nifty.) There are some really great usability improvements, though, that are only marginally related to special effects:

• Shadows behind windows make it easy to see where one window ends and another one begins. It's yet another visual cue, adapted to our already-stellar ability to interpret depth under varying lighting conditions. A cluttered desktop seems less so automatically. Big UI win, here.
  
• Transparency of moving windows. It's easy to see exactly where to place them to maximize on-screen information.
  
• Windows zoom in when created. Because I can see the animation, I never lose track of whether a web site opened a new window when I clicked on a link. (I run Firefox maximized on one of my monitors.)
  
• Scaling windows. I hit F8, and see every window in full. Nice.
  
• Switcher previews. When Alt-Tabbing, I see what's on every window.
  
• Desktop cube. This gives multiple desktops a kind of continuity and relative placement that a desktop pager could only dream of having. I actually use multiple desktops when I have this.
  
• Zoom. I have a friend whose eyesight is degrading rapidly, and he uses this a lot. It's ten times better than any kind of desktop magnifier. Also, my eyesight is great, and I like small fonts. When I want to show him something, I can hold down Super and scroll the mouse wheel up so he can read it.
  

I also do image processing research, so that last one is great when I need to see fine detail. None of these by themselves are any great reason to have an OpenGL desktop (except Zoom if you've got bad eyesight), but they make a very compelling case in the aggregate.

Re:Leopard

[njh]

Did you know that the default gnome keys are cntl-alt-arrow to move to new workspaces and cntl-shift-alt-arrow to bring the focused window with you? I jump around so fast that any animation just annoys me.

Re:Future

[Rutulian]

Uh, no, you have a couple of things mixed up. Xgl/AIGLX is the part that goes into X11 and provides the hardware accelerated 3D functions. Compiz/Beryl is a compositing window manager that actually does the effects. Every window manager has access to the 3D stuff, but they each individually have to implement their own effects. Early in the game it was attempted to separate the compositing manager from the window manager, but there were problems doing this (mostly performance, I think). So now everybody agrees that you have to integrate the two. I think the GP is right. As soon as Metacity, KWin, and whatever the XFCE WM is implement their own compositing effects, Compiz/Beryl will be an obsolete experiment. Personally, I'm holding out for Metacity. I've played around with Compiz/Beryl, and I like it, but I think it can be trimmed down quite a bit, and some major usability studies have to be done to make things like the wobbly windows less annoying.

Xinerama support

[agm]

I would love to take advantage of the eye candy a 3D accelerated desktop provides, but until I can do it with a multi-head setup it won't get used by me. The COMPOSITE extension doesn't play well with the XINERAMA extension. It's a showstopper for me.