Slashdot Mirror
Attempting To Reframe "KDE Vs. GNOME"
jammag writes "Setting aside the now tired debate about whether KDE or GNOME is the 'better' Linux desktop, Bruce Byfield compares their disparate development approaches and asks, not which desktop is subjectively better, but which developmental approach is likely to be most successful in the next few years. 'In the short term, GNOME's gradualism seems sensible. But, in the long-term, it could very well mean continuing to be dragged down by support for legacy sub-systems. It means being reduced to an imitator rather than innovator.' In contrast, 'you could say that KDE has done what's necessary and ripped the bandage off the scab. In the short term, the result has been a lot of screaming, but, in the long term, it has done what was necessary to thrive.'"
In the second paragraph, the blogger says:
s/journalists/bloggers/ and you've got this story.
My pics.
Gnome and kde are designed for different types of people, in gnome everything is typically simple and straight forward, but lacks the ability to be configured the exact way you like and is less powerfl.
KDE on the other hand, gives a lot more flexibility and power over the way you have things, but the trade off is complexity.
Both will continue to be relevant to their different markets for the foreseeable future. Even if development halted right now.(not that it would)
In my opinion, despite Gnome's incremental approach, they are still highly successive in alienating their users.
"Ripping the bandage off of the scab" is a pretty accurate description of KDE 4.
No folly is more costly than the folly of intolerant idealism. - Winston Churchill
Strange. I seem to recall the GNOME project being started because of KDE using the Qt toolkit, and then trying to catch up with KIOSlaves, DCOP, KParts and other superior technologies in the KDE camp.
War is one of the most horrible things a human can be exposed to. And one of the worlds largest industries.
From the article you get the impression that KDE use radical changes whereas Gnome strive in little steps...
How in accurate. Both evolve in little steps and both occasionally make radical changes.
Gnome had a major remake for 2.0 which reduced the older clotted layout.
KDE had a major remake for 4.0 which vectorized most of the gui.
Otherwise, changes are small. For both.
.
Reason I don't like Gnome, is because GTK simply isn't good. I mean, it can't even show a window inside a window to get MDI or floating toolbars. There are almost no complex programs with a good GUI in Linux (programs like photoshop, paint shop pro, 3ds max, ms office 2007, ...), because GTK doesn't support doing floating and dockable toolbars or multiple open files in a good way. Blender is one of the few programs with a complex well done interface in Linux, but they did the entire GUI in OpenGL I think, not using a library like GTK.
I don't know why, but this is related to the philosophy of Gnome and GTK and since I don't like that philosophy, I don't like Gnome either.
Since 4.2, KDE4 has become quite usable. I already prefer it over KDE 3.5.
The real edge of KDE over Gnome has always been the tech, though. kioslaves vs. gnomevfs is one example, KParts another. Add Qt 4.5 to this, and it becomes obvious that KDE is vastly superior under the hood. But, this is not what users are interested in. I do think that KDE4 learned a lesson or two from Gnome about this. I just hope they don't start removing all options because they think the "user may be confused" (just like with the infamous printing dialog Linus Torvalds was so frustrated about).
This sig does not contain any SCO code.
Yes, because we all know you have to throw out the baby with the bathwater every 5 years to "innovate". Uh huh.
Say what you want, but five years ago it wasn't reasonable to design the modern composite desktop. Five years is still a long time in computing and it shows. Think of it a little bit like construction work - you can remodel an existing building but if you really have to change the fundamentals you build new. That means you get all the fun of working out the kinks in the plumbing and wiring and whatnot all over again, and for a while that sucks. Then you realize it's actually quite great to live in a modern building.
Kinda seems like KDE is the imitator.. kinda seems like KDE has always been the imitator.
KDE is by default imitating a lot more, then has the configurability to decide where you want to be innovative. Desktops are very much "works for me" kind of stuff, when you like the "new way" that's great but Gnome has pulled a few on me where I just go "why couldn't you just leave this the #""#& alone and don't mess with it?!" and the way to revert it is usually in some obscure gconf option or no longer available because it's not "supposed to" function like that.
I've worked with Qt4 quite a bit and it's become a very complete and consistant toolkit. The changes were large, painful and it took quite a while to get everything working as well as in Qt3. I think the same will be true of KDE 4, once the dust settles it'll have the potential to rise much higher than KDE 3.5.10 and Gnome. As well it should, it's OS X setting the standard these days...
Live today, because you never know what tomorrow brings
You can do this just fine on Debian/Ubuntu. Choose a decent distro if you want decent installation options.
ripped the bandage off the scab
Eh.. that is usually a bad thing to do.
I just tried out the Ubuntu and Kubuntu 9.04 betas earlier today, and I think my interest in both GNOME and KDE is just about worn out.
Both are really quite bloated. I've been on Debian and KDE 3 for years, but I think I'll be switching to a stand-alone window manager like fluxbox, or maybe Xfce, the next time I have to upgrade.
GNOME on Ubuntu felt as sluggish and amateurish as ever. No amount of new themes and rehashed icons can improve GNOME. As a KDE user I was looking forward to KDE 4.2 but christ, it's so damn cluttered. I think they've actually added more clutter since 3.5, not taken it away. Every damn UI element flickers and flashes with a mouseover effect as you move around; some kind of indexing service is hitting the disk in the background; there's a plethora of desktop views or applets or whatever they're called, none of which I'm interested in; there's a new K menu that looks like it was a reject from Windows XP, and which takes several clicks to hunt around for what you're looking for; the default widget theme has super thick borders, even the pull down menus have thick borders around the menu items. The whole thing is just over-cooked. I couldn't make sense of it, frankly.
Sure, I could turn off or tweak most of that junk. But I think what I saw today is what happens when you try to copy Windows and Mac too closely. You end up copying the bad as well as the good. You inherit the same limitations and the same performance standards. It's a poor form of competition, and I despair at how much programmer effort must have gone into creating all this bloated mimicry.
Having said that, I only just scratched the surface. I know how good Qt 4 is, and I'm sure developing apps with the KDE4 framework is much nicer than KDE3. It's just that the result on the desktop (both of them) is a bit of a let down.
One of the major effects KDE 4 has had on the free desktop has been to light a fire under the metaphorical asses of Xorg and driver development. There has been tons of work going on in Xorg since the split, but until KDE 4 came along and proved that stuff like Composite could have a real effect on user experience (Compiz came first, yes, but that was more or less just bling until apps started using composite), there was not as much pressure and expectation from free desktop users.
Turn on desktop effects on any system using KDE 4 and if you have Xorg with good drivers, the difference in experience is startling.
The rate at which Xorg and some of the drivers are getting better is exciting, as is Qt and KDE itself, and this is in part due to the expectations that KDE 4 has set in the minds of free desktop users. Kudos to the Xorg and FOSS driver devs for stepping up. The next couple of years are going to be fun.
To give an example, Gnome's file browser takes 5 seconds on my home PC (Athlon, 2GHz, 3GB) to list a 161 entry directory. A virtualised W2K instance on the same box takes less than 1 second to list the same directory - even though it's running in a VM and has to go through SAMBA on the host to access the directory. When doing this, I took precautions to ensure no entries were cached on either instance.
Whether that's due to a mis-configuration on my part (tho' the Ubuntu installation is simply "out of the box", no tweaks) or because the browser is badly written and poorly designed, I don't know.
What I do know is that this effect is not limited to the file broswer and is a severe demotivator for using Linux - or recommending it to others.
Lose the bloat, remove 50% of the features, optimise the code, THEN talk about which desktop is best.
politicians are like babies' nappies: they should both be changed regularly and for the same reasons
We must have been using a different GNUStep.
As someone who has been using KDE 2001 (around KDE 2), I have to say that I think the latest version of KDE is fucking shit. It's a MAJOR step backward from KDE 3. I feel like the developers have taken everything that was good about KDE, thrown it in the bin, and made every effort to drive me to another DE altogether.
Things that have so far fucked me off:
I upgraded to KDE 4.2 a while back after everyone raved about it, but ended up reverting back down to KDE 3.5. I'm still not sure what the KDE team are attempting to achieve, but I would rather have seen a KDE 3.6 with all the fancy effects than what we have now.
I'm going to look very carefully at KDE 4.3 when that comes out, but I have little hope that it will reach the 3.5 standard, if I'm totally honest. Rant over. Sorry, had to get this off my chest. Am I the only one that feels this way? I'm sure when 4.2 came out Slashdot commentators were proclaiming it to be THE KDE 4 we'd been waiting for. Not me.
THE HONOUR OF THE KNIGHTS - CC Licensed Sci-Fi Novel
"For KDE, why won't you then customize it to your heart's desire?"
;) ).
That is not the point. Read his post - he said "I could turn off or tweak most of that junk".
If the OSS GUI people keep picking crappy defaults and require 90% of the people to customize/tweak stuff to achieve "decent usability", then that means their desktops are unsuitable for public use - it means they FAIL! Sure one may feel Windows requires lots of tweaking etc to be decent, but it has the market advantage of being "defacto/preinstalled".
A good GUI designer picks good defaults, so that 90% of the people will find it tolerable or even usable and won't need to customize it.
Think of GUI design as "user choice + huffman coding". The most popular options should be only one or two clicks/choices away, the advanced options should still be possible, just more steps.
GNOME fails the latter - they seem to have the development philosophy of totally removing/hiding features just because they might confuse the user.
KDE fails for having poor defaults. Look at their latest default menu, how many people want to keep clicking backwards and forwards to navigate their stupid new menu to look for the application to launch? BTW I tried Kubuntu recently and KDE was crashing way too often - so that's another fail.
1) The typical desktop user would not know how to customize his/her desktops OR want to know, so the desktop environment FAILS if it requires customization to achieve a good level of usability.
2) Even if there are "resident geeks" around to customize stuff for the desktop users, this results in zillions of different customizations because every geek will have their own favourite customization. This creates a big problem when users try to call 3rd party "Customer Support/Helpdesk" - the helpdesk agents and people writing the helpdesk scripts won't even know where the caller's taskbar will be.
At least with windows, the typical user's "start button" will be in the lower left hand (windows users who have moved it elsewhere don't normally call support to look for basic help- they call support to try to get to some higher level tech
How would the designer pick the defaults? They could test various designs with a large sample of users.
Just asking people what they want doesn't work that well, because often the users themselves don't know what they want or don't say it. After all, millions of people wanted chunky spaghetti sauce, but never said it in surveys till Howard Moscovitz did some taste tests with dozens (100?) variations of spaghetti sauce and found that a lot of people liked chunky sauces (at that time there were ZERO chunky sauces on the supermarket shelves!).
So a good designer will narrow down the variations (getting rid of the totally crappy ones - you don't bother testing varieties of spaghetti sauce that are totally awful) to a manageable number of varieties for testing.
Have you considered that it could be a conscious decision, because MDIs and dockable toolbars are ugly and annoying? OSX doesn't use either of those UI paradigms, and developers don't cry out for them. As a user, I find OSX's floating, contextual inspector palettes to be much nicer than the mess of toolbars and dockable crap in visual studio. I'm getting a Linux box this week, and if Gnome doesn't have MDIs, I think that's one more thing to push me in that direction. (I think I'm going to choose Fedora as my distro.)
the troll has 1 relevant thing to say here:
instead of getting your shit together.
Now, I don't care so much about gnome v kde, but I do wish there was more consistency for all Linux GUIs. If everyone had a common standard to work to (eg the Windows Style Guidelines) then the Linux desktop would become a better place to work. MS did wonders for themselves with this, and until recently kept with it - unfortunately, now they've replaced the menu bar with a round button thing, no-one can find the print option anymore - which only goes to show how important and powerful the guidelines were.
Linux has the opportunity to be great (we all know that, even the MS trolls), but isn't necessarily following up on its potential. Gnome v KDE is probably the biggest factor stopping this from happening.
Yes, indeed I have considered that it is that decision. That is exactly the philosophy I don't like that I mentioned. That is the problem: the lacking features of GTK aren't due to lack of developers and time, but due to these decisions.
I'm not convinced of the advantage of these decisions. Also you say OSX doesn't use it, but OSX is conceptually a totally different type of dektop. In Linux, how something like Gimp looks, sucks.
Also, whether or not MDI is useful might differ from person to person, but I'm sure a lot of people, including me, like it, and there's no reason to leave it out from a proper GUI and desktop.
He even calls D-bus "inspired by DCOP" but ignores the fact that D-bus is not part of gnome, but gnome has instead switched to a universal standard that is not desktop-specific, and was already used by non-gnome applications on the system, including low-level components such as udev and hal. I'd wish KDE would do the same, no one needs a 2nd seperate message bus system on his machine.
...which is why, probably, KDE4 uses DBus. But don't let the facts get in the way of a good rant.
Slackware moved KDE4 from testing into -current only a few weeks ago. So I was expecting that it is considered ready for general use. I was disappointed to find out, that the major applications such as KDevelop, Quanta and K3B are missing. And they will not come out soon either. KDE 4.2.2 will be released in a few days and still it will not contain KDevelop/Quanta/K3B. There are no dates given beyond KDE 4.2.2. KDE 4.0 was released in January 2008 (with alfa and beta releases published months before that). A year later the major apps are not ported. The change is too drastic if the major applications can't catch up in reasonable timeframe.
The first thing which was really annoying is, that the "Dolphin" file manager eats about 200mb of ram almost instantly.
Hi. Many things have changed since KDE 4.1. Over here (running KDE 4.2.1), Dolphin has a virtual size of ~71MB and a resident working set of ~20MB. You might want to look into upgrading.
$ ps -eo vsize,rss,comm | grep dolphin | grep -v grep
71652 20868 dolphin
$ dolphin --version
Qt: 4.5.0
KDE: 4.2.1 (KDE 4.2.1)
Dolphin: 1.2.1
Ah. Something that I just thought of... is your version of KDE an optimized build? (I'm not sure that this would make *very* much difference at all, but...) Over in my full debug version of KDE SVN trunk, Dolphin has a virtual size of ~128MB and a resident size of ~28MB.
$ ps -eo vsize,rss,comm | grep dolphin | grep -v grep
128268 28532 dolphin
$ dolphin --version
Qt: 4.5.0
KDE: 4.2.68 (KDE 4.2.68 (KDE 4.3 >= 20090327))
Dolphin: 1.2.80
The floating toolbars in osx have a load of problems, including constantly being hidden in the background somewhere when you need them because they do not stay on top. The reason osx never has used mdi simply was it is not really possible in osx in its metaphor of having the toolbar on top, every osx application sort of is mdi but the mdi is the desktop. I am not sure if this is the correct approach, my guess is either approach has its fair share of problems!
Reason I don't like Gnome, is because GTK simply isn't good. I mean, it can't even show a window inside a window to get MDI
Sup dawg, I heard you liked windows so I... nevermind.
Give me Classic Slashdot or give me death!
x.org comes to mind, do away with the client/server paradigm,among many other things...
Because taking away one of X's greatest features is going to make it better how?
This misconception comes up all the time. Applications need to communicate somehow, you might as well use Unix Domain Sockets. Once you have a protocol that runs over unix sockets, you can just as easily send it over TCP sockets. Network transparency comes essentially for free as a side effect of designing the local IPC efficiently.
Do you have a situation where you can demonstrate that the network transparency of X is a bottleneck? People have been harping on this issue for years, and no one has ever demonstrated such a bottleneck. There's no reason to believe it is a bottleneck at all.
Give me Classic Slashdot or give me death!
I get the feeling he meant Qt or something.
"Thanks for all the money you paid to us. We've used it to buy off ISO among other things" -Microsoft
No one gives a shit about network transparency. It is a totally meaningless buzzword engineers like to tout but doesn't mean anything in the real world. X is mostly network transparent, X clients aren't. You have to be very careful to avoid unnecessary round-trips which introduces latency and makes your application dog slow. The Windows shell is decidedly not network transparent, but RemoteDesktop and VNC still works with it.
To understand why people are complaining about X, try resizing a window quickly. Do the same operation on windows. It doesn't matter what computer you are using, on X you get flicker. Try opening a bunch of apps on one workspace, then move away and to that workspace. Notice how each window is redrawn one by one, first the frame and then the window contents. That is also an effect of X's client-server architecture. If you use some other OS than Linux so you have something to compare with, it is easy to understand why people complain on X.
Football Odds
Usability is inversely proportional to the number of mouse clicks required for the user desired feature.
I'd like to buy homeland for our 10 million people. http://twitter.com/mahadiga