2010-10-11

GNOME 3, Activites, and KDE 4

There have been a slew of new articles detailing the progress of work on GNOME 3, and the refrain in all of them has been that "GNOME 3 will revolutionize the desktop". The focus on GNOME 3, ever since the release of the first mock-ups, has been on the new GNOME Shell and GNOME Activities (which are really just two sides of the same coin). The thing is, GNOME Activities has essentially the same concept (and even the same name) as KDE 4 Activities. So I was thinking for quite a while: how can this be called "revolutionary" with a straight face? Today it hit me: while KDE may have had the idea first, GNOME presents a far superior execution of this idea; GNOME Activities in the alpha and beta versions of GNOME 3 was very usable and improved with each iteration, while KDE Activities remained very slow, very buggy, and nearly unusable until the release of KDE 4.5.
All this makes me rethink my previous position on GNOME 3. I previously believed that GNOME 3 would suffer the same fate as KDE 4, in that a lot of current GNOME users would migrate to other DEs upon seeing GNOME 3 (be it for its radical nature or its buggy nature). Now, however, I don't think this is the case. I think the major *nix DEs are finally falling into fairly well-defined niches. GNOME will emphasize simplicity, ease-of-use, and understated modernity over flashiness and over-the-top effects. KDE will be the way forward for ultimate customization, web-connected computing through Plasmoid widgets, and flashy desktop effects (as well as tools for power-users, like Dolphin/Konqueror vs. Nautilus, Okular vs. Evince, Kate vs. Gedit, etc.). (Xfce and LXDE will, of course, remain the DEs of choice for people who need lower-resource but still fully-functional and modern DEs.)
But with GNOME moving towards a more tightly-integrated and powerful Metacity WM, one WM is still left out in all this: Compiz. Unfortunately, Compiz and its desktop effects still don't work in recent builds of GNOME 3. While Compiz integration with KDE has gotten better, it still isn't seamless, and Kwin is almost there (but not quite). While most everyday Linux users don't use most Compiz effects (except maybe window decoration transparency and minimize/maximize effects), these effects often play a role in convincing non-Linux users to try Linux. There have been stories after stories of people just using their Linux computers with their friends and their friends being awed and intrigued by the desktop cube and the wobbly windows; don't underestimate the power of these effects to convince people (in the implicit form of "can your OS do this?"). So what does all this mean? It'll become a lot harder to convince people to use Linux through this route, as there will be many people put off by the confusing and endless customization options of KDE 4 (or simply can't run it because they have lower-end hardware). So, GNOME 3 developers, can we please get Compiz integration with GNOME 3 before the first official release? Thanks!

16 comments:

  1. You shouldn't worry about Compiz. Mutter will provide the desktop effects. If you really, really want Compiz integration with GNOME 3 you are out of luck. Don't ask me why but GNOME developers designed GNOME Shell to be a Mutter plug-in, so as you can see the former depends heavily on the latter, thus making impossible for Compiz developers to support GNOME Shell.

    ReplyDelete
  2. @Anonymous: If Mutter can provide feature-parity (along with the same stability and speed) with Compiz for the major effects (e.g. shift switcher, desktop cube, Expo), I'm OK with that. My problem is that right now, the most Metacity can do as far as effects go is probably transparent window borders. My concern is that's all Mutter can do as well (without Compiz); if it can do most of the big things that Compiz does as well as Compiz does them, I'm fine with that. Thanks for the information!

    ReplyDelete
  3. If I recall correctly I believe I once heard Compiz was never supposed to be permanent. It was an example of what the Windows managers (aka GNOME an KDE) could and perhaps should/should not do.

    ReplyDelete
  4. @Anonymous: Really? I wouldn't be surprised to know that Compiz was originally just a proof-of-concept sort of thing, but I would be surprised if the developers still thought of it that way. I mean, it's an incredibly fast and versatile window manager in its own right, and even without the desktop effects, it can be used instead of, say, Metacity or Openbox. Thanks for the comment!

    ReplyDelete
  5. A lightweight desktop with modern features is E16. It has transparency and excellent 2D desktop management. E16's clear distintion between virtual screens and virtual desktops implemented the concept of "activities" more than a decade ago. Got a new "activity"? Just create a new desktop with as many virtual screens as you want and open the applications in a way that works well for you. Want a modern panel? Run the 3.5 branch of KDE's kicker or the gnome-panel, both work well and can be stuck into position.

    ReplyDelete
  6. @twitter: It sounds intriguing, but hasn't E16 been mostly superseded by E17? Also, I'm looking more at prepackaged fully-fledged DEs, not just WMs. Thanks for the comment!

    ReplyDelete
  7. "Today it hit me: while KDE may have had the idea first, GNOME presents a far superior execution of this idea; GNOME Activities in the alpha and beta versions of GNOME 3 was very usable and improved with each iteration, while KDE Activities remained very slow, very buggy, and nearly unusable until the release of KDE 4.5."

    Wishfull thinking. KDE activities are working excellent in KDE 4.5 and this is a stable feature. Gnome 3 will be released next year if everything will go as supposed, so ask yourself how is this possible to compare something which works right now (KDE 4 activities) to something which will start working as supposed next year? Gnome 3 is in unusable state right now, its activities aren't revolutionary in any aspect (read A. Seigo post). It's Gnome 3 which is damn slow right now, it crashes right now, it's far from being feature rich right now, so I don't get the idea of your post. Btw. KDE 4 activities are far more advanced (right now (tm)) then 'activities' which will be delivered with Gnome 3 next year.

    ReplyDelete
  8. And of course, GNOME's implementation of Activities look somewhat nice when compared to earlier versions of KDE's: KDE ALREADY went through all the trouble of figuring out its usability quirks iteratively drawing a blue print good enough for GNOME developers to know before hand what are the pitfalls, what to do and what not to do. Why not compare it with the KDE SC 4.5 one?

    Furthermore, it is incredibly stupid to compare something that still is in the pipelines to something that works right now and say that the upcoming one "will" be better because it "will" do this or not do that!

    I can't shake off this feeling that no matter how better KDE tech might be compared to GNOME - having a superior underlying GUI toolkit and all - and that it has been laying down the ground work for all the FOSS *nix desktops to follow, certain people refuse to acknowledge it or even give it the credit that it deserves...

    ReplyDelete
  9. @Anonymous: Are you specifically referencing me or other people? I do credit KDE 4 for first introducing the concept of Activities, and I do mention that with KDE 4.5, Activities are finally usable and stable. Also, while it is true that GNOME 3 is still technically "in the pipelines", GNOME Shell and Activities can be used by anyone using GNOME 2.X (probably 2.28 and above), so it's a current technology (meant to preview a future environment). I don't disagree that KDE is groundbreaking and ahead of the curve, but what I'm saying is this comes at the cost of stability, while GNOME typically waits a few release cycles to iron out these bugs because stability is more important than cutting-edge features. Thanks for the comment!

    ReplyDelete
  10. I was surprised to find out that Compiz still exists. Kwin, Fluxbox, and Metacity have all, to some degree, incorporated this. I know they aren't as flashy as compiz, but I think it's just a matter of time. Compiz was the fire under the butts of developers, showing them what X could do and daring them to match it. I was very into Compiz for a year or two. But, since I do high-end stuff like 3D rendering, it was just slowing my computer to have it enabled without providing any benefit.

    I have to say that, in my experience, everyone who saw Compiz thought it was neat, but no one was converted because of compiz. They wanted to know if they could still do the work they did on their windows computers. I think it's mostly young kids or something like that who actually care about those dumb graphics. The utility of having multiple desktops (copied by OSX) is more important than whether it's represented on a cube or just slides into place.

    ReplyDelete
  11. @Eric Mesa: Fluxbox? Are you sure? I was under the impression that Fluxbox, like Openbox, is a highly-configurable, minimal effects, no-frills WM, but then again, I'm probably wrong. Anyway, it's always nice to play around with the cool effects, but ultimately, in my experience as well, it just needlessly slows my computer down (though this is less of an issue now because I have more RAM and video memory than I ever anticipate needing). Thanks for the comment!

    ReplyDelete
  12. "The thing is, GNOME Activities has essentially the same concept (and even the same name) as KDE 4 Activities."

    I guess you didn't read Aaron's blog post, which does a pretty good job of describing how, in spite of the fact they used the same name, they are not the same concept.

    http://aseigo.blogspot.com/2010/10/activities-as-homonyms.html

    ReplyDelete
  13. @Anonymous: Yeah, I read that, but unfortunately it was after I had already published this post. Yes, I wrote in error. However, after rereading the article, what I will say is that unless specific applications (and not just Plasmoids) operate differently in different Activities in KDE 4, that one very popular thing to do (though not the default behavior) in KDE 4 is to tie different and new virtual desktops to different and new Activities, and that considering GNOME 3 doesn't have Plasmoids anyway, visually, Activities operate much the same way in GNOME 3 as they would in KDE 4 if Plasmoids were the same across all Activities. This, of course, isn't really saying a whole lot, but do consider that both involve zooming out and presenting all the windows in each virtual desktop. I hope that makes more sense. Thanks for the comment!

    ReplyDelete
  14. "considering GNOME 3 doesn't have Plasmoids anyway, visually, Activities operate much the same way in GNOME 3 as they would in KDE 4 if Plasmoids were the same across all Activities."

    In that description you would just be using multiple desktops, that's totally ignoring what an activity *is* in KDE 4.

    I can create new activties in KDE like office, video editing, etc... customize each desktop relative to different tasks related to that activity, if I switch to the desktop activity, all the customizations on all the desktops are as they were the last time I was running that activity, if I switch to the office activity all the desktops are customized that way they were the last time I switched to that activity, if I switch the the desktop activity all of the desktops are customized the way they were the last time I ran that activity.

    Later, Seeker

    ReplyDelete
  15. @Anonymous: You are absolutely right. However, as I do not use Activities to their fullest extent in KDE 4 (because I haven't really seen the need for so much organization), in my daily usage the distinction is meaningless. Thanks for the comment!

    ReplyDelete
  16. "in my daily usage the distinction is meaningless."

    Activities in KDE may be meaningless to your daily usage, I accept that.

    If you limit yourself to comparing multiple desktop usage in Gnome-Shell to multiple desktop usage in KDE, without saying you are comparing activities to activities, the fact that activities are completely different things becomes irrelevant as well.

    But when you write something that gives the impression that you are comparing activities in KDE to Activities in Gnome-Shell, the distinction is important because they are completely different things.

    This next part is all speculation, but the decision to rename desktops activities would seem to imply intent to build on that idea by allowing the saving and loading of customized activities and it seems highly likely that an activity will continue to be limited to a single desktop.

    If that does actually happen, then the definition of activity becomes even more important. A single desktop in KDE would still just be a single desktop, but in Gnome it would be something more. Even comparing activities may not seem all that relevant since in Gnome you could load multiple activities but they would be confined to a single desktop and in KDE you could define activities that span multiple desktops but you can only load one activity at a time.

    Of course speculation may be irrelevant since we don't *know* how the KDE/Gnome desktop environments will evolve and they each have the option to implement ideas from the other.

    Just something to think about for the next time.

    Later, Seeker

    ReplyDelete