Nicholas Petreley installs and compares the two most popular desktop environments for Linux & other Unixes. His recommendation? It depends… Update: OSNews reader Cesar Cardoso just sent us this: “Why not a UI-fight between Havoc and Mosfet?” This is Mosfet’s rebuttal article answering to Havoc Pennington’s article, while the first part is here.
A Brief Comparison of KDE 3.1 and GNOME 2.2
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
I love Gnome2 so far, but it’s pain in ass sometime. There are too many different dependencies than KDE has. I never understand why can’t Gnome2 just glue them together. I don’t like gconf and too many ~/.g*, but I don’t bother them thought..
Both of them seem rather clumsy. The null/Bluecurve initiative does a great deal for making things look and feel better, but I don’t think things are just quite there yet. For the moment, I’d sooner run something like Enlightenment.
‘Desktop Enviroments’ unless he means kwm and metacity
KDE:
Consistant
Nice eye candy
Most apps are more polished than Gnome counterparts
Advanced features are integrated faster
More commercial support
GNOME:
Not everything has to be rounded and bubbly. Thats nice.
The foot icon owns that retarded K
Actually with the improvements in terms of options to the interface in Gnome, I see only three desktop things I find truly aggravating with Gnome.
1. Menu editing needs to be improved. The Nautilus menu editing is annoying and there is no real tie in or indication of its use associated to the Gnome panel Applications menu. In other words, in the menu, there is no way to launch the menu editor.
2. The mislabeling of the real Gnome Control Center. Despite what some think if you use RH8, you have a real Gnome Control Center. It is just not labeled at all. I am talking about the start-here icon on the desktop. If you go to the command-line and launch the gnome control center, you get the desktop preferences. That is not what people think of when they think of a control center. The start-here menu in RH is. What do I mean? There are entries for Application menu editing, Server settings, System Settings and Desktop Preferences. Only RH takes full advantage of this but it is there. Come on, even in SuSE, it takes me about 15 minutes to copy the Yast2 shortcuts into place to emulate this behavior for start-here.
3. Consistency. Deal is that I tried the same trick the author did with Gnome2 versions of Galeon and Gnumeric and Nautilus. I got no right click context menu for the task bar on Nautilus and Galeon. I did get one for Gnumeric. There are some projects that refuse to follow the HIG or slow to make these changes. If right click options on common sets of interfaces have no correlation from app to app then that will hurt. The funny part is that I prefer the look and feel of most Gnome apps, gtk apps to QT/KDE apps any day of the week. I started giving Gnome a chance because I realized that the vast majority of my apps were gtk/gnome apps. Once you get Dia, Gnucash and Evolution over to the GTK2 side perhaps things will tighten up around the HIG. I hope so.
Yes, Nicholas Petreley actually finding something good in GNOME… Soooo different from his old “GNOME is crap and everybody should go KDE”. The GNOME folks should have done something good on GNOME 2
once there, they face all the troubles Windows has to deal with before
How much do (or will) KDE 3.1 and Gnome 2.2 change across distros? I mean, how much different is (or how much different will) KDE 3.1 in Suse be vs Mandrake vs Slackware, etc. Similarly, is Gnome 2.2 in the new Redhat beta radically different than whatever implementation of Gnome 2.2 that Gentoo has?
Also, are there any really good (read: plain English) docs that explain the different parts og KDE and Gnome (what the fuck is Bonobo anyway), how they work together, how to compile and install the things from scratch (I hate waiting for binaries), and basically how they integrate/work together with X (by that I mean, what is the distro-indepdendent way to tell which DE will launch when you type startx)? Also, how to run one version of KDE/Gnome along side another?
I don’t have much interest in Linux apps (yet), but I’m very curious to learn how all of this stuff works .. I’m in the tinkering mood
gentoo uses gnome 2.2 straight from the source. in this case the gnome 2.2 ftp server (or atleast that’s what my setup does) and on my ibook I’m running a buildscript version of gnome-cvs called garnome.
Gnome and gdm installed perfectly with Debian unstable on my box. I really like GNOME’s simplicity in look-and-feel, but that’s about it. There are SERIOUS usability problems in GNOME at the moment, and KDE is kicking butt.
is kicking no butts…
kde is good if your into integration.
off the top of my head:
konqueror: whilst playing music and perhaps doing graphics at the same time and then try to access a website it usually “breaks the connection” since it can’t bind the sockets required same with kmail and other web related applications.
kmail has some serious bugs and usability issues aswell.
the kicker has a imo cluttered application menu, too many categories, recategorise them into subcategories.
the menu inconsistencies of gnome are being worked at, and galeon/-2 is not a gnome only browser it integrates nicely with gnome thats all, epiphany is however the new gnome lovechild.
i’ll settle for that both have issues and both are being adressed.
If only KDE and GNOME could use some idea’s from each other and borrow some more things, like maybe both can use the same skins and themes and stuff. If both were to help each other with the things they dont really accel at they could become more competitive. I dont think this would ever happen because us open-sourcers need competitiveness to progress faster.
How much do (or will) KDE 3.1 and Gnome 2.2 change across distros?
Depends on the distro. Gnome is pretty much the same across all the distros except Redhat because Redhat is the only distro to tweak Gnome from the default much. SuSE and Lindows and loads of other distros do minor tweaking here and there to parts of KDE. They vary in minor ways mostly cosmetic except Redhat that tries to make both environment work the same and therefore KDE folks say that RH cripples KDE (other reasons too).
Similarly, is Gnome 2.2 in the new Redhat beta radically different than whatever implementation of Gnome 2.2 that Gentoo has?
All the major stuff is the same but Redhat took the time to have place a mime type that works off of a doubleclick to install an rpm for example. The menus are better organized. The icons are nicer and it comes with a pretty theme. If you go into Start-Here on the Desktop and into System Settings all the redhat-config-tools are there.
Also, are there any really good (read: plain English) docs that explain the different parts
I do not know about KDE but here is the gnome link:
http://developer.gnome.org/
http://developer.gnome.org/arch/
I don’t use KDE.
Also, how to run one version of KDE/Gnome along side another?
For SuSE and a number of other distros put kde2 in /opt/kde2 and gnome in /opt/gnome and gnome2 in /opt/gnome2. Having versions sitting side by side (I did this with SuSE 8) is not really that god awful. With Gnome, there is another way to stay on that bleeding razor’s edge and that is garnome.
I love Gnome and it has come a long way.
Still, KDE is a fine desktop I just personally find it bit overloaded and bloated but that is soley my opinion and I hate the look and feel of KDE apps once again a very personal thing.
Good tinkering Darius.
I will not say I hope you end up using linux. I only hope you have fun tinkering and continue to use the OS that is best for you.
And I wonder why he tries to install the packages on Debian UNstable while there are KDE packages for Debian stable available which are additionally complete (including kdenetwork and kdepim). I guess the same problem existed for him with the Gnome packages.
I haven’t been able to find a desktop environment that I like. I don’t like the inconsistencies in gnome/administration problems. I don’t like gtk because a c-based toolkit is very behind the times and much less useful to me (yes I know about the c++ wrappers…but that’s one more library to worry about). Also I think nautilus is way to slow. I run on a slow computer and nautilus is almost unusable. If all I want to do is quickly look something up on the filesystem it’s faster to open a console and type my way there than it is just to open nautilus. That said, I don’t care for kde either. Administration-wise, it’s much better. I also appreciate the unified look and feel and layout. I like the eye candy. I also like that konqueror is faster than nautilus (form me). I just don’t like the clutter. konqueror has way too many buttons. They don’t need to be there all the time. The button set should dynamically change based on your task. Maybe there’s a way to do this, but I missed it, and besides shouldn’t simpler be the default with the ability to add more complication yourself? KDE is just too messy. On top of that I don’t like that QT uses it’s own extensions to the C++ language. It seems like it should be doable with standard ISO/ANSI compliant C++ to make a gui toolkit. That’s just my say on the matter.
From Havoc’s article: [Admin Edit: The following was written by Mosfet, not by Havoc.]
“Listening to your user base and taking into consideration what they say they want to see is a much better one. Believe me, there is no lack of feedback from users about what they want ;-)”
He’s making two assumptions here:
1) the feedback you get from the users is representative
2) users know what’s good for them
Let’s pick the second first:
Do users know what’s good for them? If so, why do we need experts at all?
Let’s exaggerate a little: If people knew what’s good for them, why not leave them the option of selecting the time slice size in their OS (without editing kernel source). I’m pretty sure lots of people would go for really short time slices because everything feels so snappy then – they don’t even know what context switch overhead means. People konw what’s good for them? No need for nutrition experts, just eat whatever you want. Cheetos all day long? You sure know what’s good for you.
Am I going to far? Yes, I said I was going to exaggerate. However, you should first make sure your users do know what’s good for them before assuming it. No one assumes the user is an expert in technical things, that’s why engineers and programmers make decisions. Why do you assume everyone’s a UI expert?
Assumption No.1: The feed back you get is representative.
I think this assumption is wrong.
First, probably a huge number of users that think KDE is flawed, do not mail the developers – they just don’t use KDE.
Second, some people are more vocal than others. Did you read that recent interview with Bill Gates that only 1% of the support calls Microsoft gets are bug reports? Be honest – when you encounter a bug, an inconsistency, a slight problem in a program – do you report it to the developers?
Finally, as a user you are not even aware of many usability problems. Often, users blame themselves instead of the UI designer when something goes wrong (“I’m too stupid to use this!”) and don’t notice where the problem is. Ever tried to push a door when it needed to be pulled? It’s not you being stupid, it’s just that someone designed a door that does not give enough visual clues in which direction it opens. (here a nice book recommendation: If you have any interest in usability at all, Don Norman’s “The design of everyday things’ is a must-read.)
> I don’t like gtk because a c-based toolkit is very behind the times and much less useful to me [..] I don’t like that QT uses it’s own extensions to the C++ language.
Choosing a desktop environment based on such criteria is ridiculous.
>Choosing a desktop environment based on such criteria is ridiculous.<
I think it’s not the best criteria, but object orientation is all over UI design. I’m no expert, I’ll admit, but I’d be willing to say that C++ is probably less tedious / error prone than C when it comes to something that OOP was born for. That has to do with a sort of idealogical consistancy, too- designing something like a UI with OOP inhearant in the programming language you use.
I think the operative phrase is “taking into consideration“. That’s just generally good sense, and not the same as putting the user in the drivers seat. Any good creative person develops the ability to listen including UI designers. Add all the rest that makes a good UI designer, and you’re certainly closer to your goal, than the one who’s going “Lah,lah,lah,I can’t hear you”.
There’s some really good discussion about the Mosfet/Havoc debate at pclinuxonline.com:
http://www.pclinuxonline.com/modules.php?name=News&file=article&sid…
Actually, the debate was originally between Eugenia and Mosfet. Mosfet wrote the first article in response to Eugenia’s comments here a few days ago:
http://www.osnews.com/comment.php?news_id=2841#74549
Later, Eugenia pointed out Havoc’s article and a few days larer, Mosfet then wrote the rebuttal, responding to Havoc’s article.
Basically, Eugenia agrees with Havoc on most points and adds some points of her own, but Mosfet doesn’t seem to agree with (regarding full customization and what not).
Actually it was mosfet who said that
Thanks for the edit. My bad.
Quote:
“From Havoc’s article:
Listening to your user base and taking into consideration what they say they want to see is a much better one. Believe me, there is no lack of feedback from users about what they want 😉 ”
First of all you are confusing Havoc with Mosfet-Mosfet wrote that sentence, not Havoc.
You then go on to ask the rhetorical question: Do users know what’s good for them ? – You could not ask a user this question, and with this attitude why would you want to ? At best you could ask a non-user (whatever that is)- but really who could be a better arbiter of what is good for them, than they themselves ? After all, the benevolent developer chuckling along, it is *for* them. Whatever features/options/preference do get implement will be at the discretion of those who write the code- user feedback means that the developers recieve a dirth of possible things which they could implement and mulitple ways of implementing those things-from this large pool of possible improvements the developers can and do then choose which ones and in which way to implement. Of course “users” frequently do not know the relative advantages and disadvanatges of implementing one thing in one way vs. another thing perhpas in another way. Yet in the linux world there is a much higher degree of coincidence between “users” and “developers”, beacuse all developers are also users and *nix users are on average far more technically adavanced(in terms of programming skills and or comprehension of programmability issues) than users of any other OS’s. In most corporations you have an IT department-and they remain isolated from the rest of the corporate microcosm- when management wants to implement a feature- they have zero knowledge of what can and can not be done, and what the consequences are if one feature must be implemented which precludes the ability to change things later, without having to rewrite the whole damned thing. On the same token the IT people tend to have no clue as to how the secretaries and stock managers work in their day to day routine, and they end up pulling their hair out because those damned IT bafoons didn’t take into account that if each time I have to enter an order I have to press 40 mouse clicks that the likelyhood of my making a mistake is muliplied 500 times. Linux, as such, will never be interesting for the “user” where “user” means a) has zero technical knowledge b) has no interest in acquiring said knowledge c) is unwilling to invest time and energy to acquire some technical knowledge- at best linux, as a generic computing system, can become something which such “users” can use without knowing what they are doing, but at that point the type/name/label/brand of “machine” the “users” are using, is, at least for them, moot. In office america/europe/asia etc. the operating system which is being used by average corporate droids is utterly irrelevant -for them-for they have no say whatsoever….To want “experts” in cutting edge fields, where the the pace of technological change is rapid and where the markets are constantly re-adapting and restructuring is simply absurd. THERE ARE NO EXPERTS IN THE FIELD OF UI-there are inviduals who have spent more time and energy comparing the pros and cons of various approaches and their feedback is welcome and needed but we are not dealing with a historical trade handed down from generation to generation. If I wanted to learn the best way to shape and mold glass I owuld seek a handworker that had been trained by a master craftsman, who himself had been trained by a master craftsman and so on. The fact that some universities have offered courses in UI theory does not mean that those who have graduated with degrees in these subjects are automatically “experts”- they are experts in one sense, in the sense that they have an expert understanding of the now-current theories, which for the most part are post-hoc attempts to account for what has already been done and held to have been a sucess. Most of what I do with my computer today was unthinkable 10 years ago-no one on the planet earth was capable of doing it- and what I do is so “normal” nowadays that its not even worth mentioning-you know like burning CD’s, making MP3’s, watching DVD’s, surfing the internet. When so much can change inside of 10 years, effectively throwing that which was common sense 10 years ago out the window, how can one pretend to be so “on top” of things today to believe that what they are spouting off today is going to have much validity in 10 years from now.
In the linux world there is little interest in “representative” user feedback. Linux is grassroots. Grassroots democracy means bringing the decision making to the people-not taking it away from them. Linux means bringing the users into the development paradigm, enabling them to contribute, in many, many ways- as opposed to systematically shutting them out unless they are willing to pay huge sums of money for development tools, which is the case for most other OS’s. For “representative” user feedback to work you would have to have a system where only those issues complained about by at least 5% of the total user base would be taken into consideration and evaluation by the developers- this is at total odds with the whole open sources software development paradigm, and more importantly its spirit. This would also mean mean that I would have to refuse submitted patches and improvements from potential contributors until 5% of the total userbase inisted on the same changes-well there goes 99.9% of my potential contributors….In the open source world self-selection is the only feedback control mechanism which can work- heavy handed top-down approaches where decisions are made for users result in user backlash and protest. GNOME2 has danced a fine line with regards to this, effectively alienating a sizeable chunk of their previously very loyal user base- yet they have one many knew converts too, ie. me. In my opinion if one takes what Havoc says AND what Mosfet says one comes closer to a sensible approach-sometimes less is more, eliminating preferences just to eliminate preferences, particularly if they are “prefered” preferences is stupid, GUI applications are not written for exactly one purpose-how many different uses have users found for MS Word, or Mplayer for that matter. However GUI applications are not and should not be in and of themselves infinitely flexible-but instead they should be infinitely combinable with other applications in ways that no one ever anticipated revealing new things which can be done creating new demands where none existed before.
Most linux GUI problems stem from misunderstood and misappropriated metaphors- the metaphors which governs the CLI is inapplicable to the GUI space, yet without a bridge between the CLI metaphor and the GUI space the GUI space becomes alienated and sperated from the system upon which it is built- at the CLI level single task, simple, small tools are the greatest thing in the world- at the GUI level such applications are a bane, and lead to cluttered menus and confusing/overwhelming superfluity. The MS Windows metaphor cannot be carried over 1:1 to the linux world-because beneath the GUI in MS lies *nothing* and beneath the GUI in linux lies *power*. The same hold for MacOS prior to OS X.
Well I recently switched back to GNOME 2.2 from months of using KDE 3.1 and I already start to regret my decision. When I started with Linux in early times I mostly used to develop and hype the Motif/CDE Widgetset and Environment but after a long time realized that I needed to go to something fresh such as GNOME or KDE. I then decided to switch to GNOME during these days but the applications and the environment didin’t look that cool in the beginning. KDE was a nightmare during these days as well so I quickly switched back to Motif/CDE but I kept using GTK+ and GNOME applications. After some more years I got fedup with Motif/CDE and finally switched to GNOME because I had the same initial prejudices about KDE as many other people had. GNOME 1.x was quite cool these days. There also wasn’t big differences between KDE and GNOME. There was a Control Center, there was the Standard applications such as Calculator, PIM, Emailer. Same for KDE these days, so there wasn’t a big issue which one to take. They where nearly similar. Over the time KDE was working on their 3.0 version and GNOME was busy moving all it’s stuff to a new architecture and cleaned up code and the result was GNOME 2.0. I was an advocate of GNOME these days and used the word GNOME on all my paths and in all my discussions and conversations with people. But with the 2.0 release a big part of my opinion totally changed. GNOME 2.0 was the first sign that GNOME leads into a total different direction. The look, the feel, the usability of GNOME went to an unacceptable level. Some things definately went better but some other things are the most retarded implementation i’ve ever seen. Therefore I totally agree to the Petreley guy. One day I decided myself to get my prejudices over and install KDE 3.0 and I was totally impressed with it. The consistency of the applications, the interoperability, the cleanless the only thing I didn’t liked was the way the includefiles, libraries and other stuff got placed during installation phase. All includes are stuffed in the root of the include path, the same for libraries. That’s something I totally hate under KDE but I don’t mount my comments on the things installed I try to see the System as a whole and KDE was damn good not to say far superior over what GNOME 2.0 was. Then I got demotivated using KDE 3.0 because I wasn’t able to programm c++ so I decided to switch back to GNOME (which moved on to 2.2) but GNOME didn’t satisfied me so I switched back to KDE and used it till 3.1 came out. There is a big difference between KDE and GNOME these days and the GAP is dramatically increasing. KDE is moving forward with big steps while GNOME stays in the same sauce. Little things happens in GNOME but they get addressed as BIG enchancement in the public. Now I’m back on GNOME 2.2 for some weeks because of my own development. I choosed to go back because I can spent my time into development of some unworthy stuff. But GNOME 2.2 is in my opinion not better than 2.0 and in no way better than KDE 3.1 saying this is quite wrong and don’t reflect the reality. Some weeks ago I came up with the Gnome UI review which got announced here on OSNEWS.com describing some of the major issues I see within GNOME UI’s that NEVER ever get fixed to an esthetical consistent look.
http://www.osnews.com/story.php?news_id=2706
You can read here. I talked about this issue with the gnome germany people and all of them agreed with me but getting the developers convinced is an impossible task. Now the german team is making the same mistakes that I tried to avoid each of them started to work on making these dialog consistent but none of them talk to each other and the result will be the same again, one is consistent, one more consistent and the last dialog made by someone else nearly consistent to HIG, so I stopped arguing with them about this issue since I realized it is a lost battle. Even the primary people that I have written to on gnome-developer mailinglist played the issue down as if this isn’t a problem at all. Specially their release coordinator guru who never ever contributed ONE patch for ONE application suggested me to sit down and fix all this on my own, while my primary intent was to get a team of people together who does this professional. It’s not easy in GNOME since it uses hardcoded GUI, GLADE GUI and BONOBO, some of these aspects got discussed on the same mailinglist too and these functions are being wrapped (as GNOME seem to get more and more wrapper functions these days to replace faulty and wrong planned code) with clean ones. Until someone decides them to be faulty again so they get wrappers too.
KDE is a real big project I would say it’s probably 3-4 times as big and complex as GNOME but the people there seem to stand behind that project like one TREE, they managed it to get a team of people together that work on the UI stuff but their way doing UI is imo far supperior over the way GNOME does it right now. It’s based on the old way Motif did UI while people could waste their time in coding the utility UI people can tweak the *.ui files and they then get embedded into the apps. So it’s easy for a good team of individuals to work on consistency, pixel exact layout of stuff etc. Once again things that I think are thought far better.
Some days ago there came the Filemanager Nautilus OO shittalk up on the mailinglists again and I was totally horrified by the idea regardless the fact that this kind of decision and conversation was stopped and people and maintainer agreed that they like the Filemanager to be the same way as every other Filemanager has to be. But no even yet people try to convince that the other thing is better. Well I personally don’t care what Nautilus should become at the final end but isn’t GNOME at the end of it’s planning phase already so that it should start fixing things that got planned so they could make a cool corporate desktop ? I mean GNOME is being presented on manny newspages in the business and and and. but how should GNOME mature to some sort of serious desktop if they start over with the plannings from scratch ?
http://mail.gnome.org/archives/desktop-devel-list/2003-February/msg…
After I read this message I went into the GNOME channel and asked a simple question why this OO conversation was brought up again and I expressed my worries that GNOME will never come out of the development phase and that it may scare the people (and customers) away. The reply from the release coordinator (the guy with the nice attitude who replied here to Eugenia during the GNOME 2.2 announcement) told me that GNOME will never come out of this day and that it will always be an implementation plattform (something like this I can’t exactly recall what he said but he simply said that it won’t mature). I replied to him that saying this isn’t good and his reply was ‘you are stupid’. So far for the nice conversation and all I was up is saying that GNOME should focus on the way it has planned and not do other things in the meanwhile and topple decisions that got decided already since it doesn’t make any progress that way.
I’m well aware that GNOME has some professional Software engineers working on it but most of the contributors and developers are only volunteers who work in their spare time, who does a primary other work and then when they come at home work on open source. This is a big disadvantage in open source I think. You can’t expect professional applications specially in a team of unprofessionals working on a big project such as this. Those who are professionals, who have an eye of esthetical design and good project plannings follow their route as is but those who are not professionals hardly be able to keep up.
I sometimes wonder how the KDE people manage all this but I think they have a lot of people backed and the value of professionals outweights the unprofessional ones or back these who need help.
To say, If you need a desktop for business and normal real work where you don’t spent half day configuring the printer, the sound, getting video to play or get killed with cryptic configuration systems or where you expect the tools to be present that you get your work seriously done and where you expect total integration, consistency and other stuff. Then my answer as hard it sounds is ‘use Windows or MacOS’ you like to stay on Linux ? then use ‘KDE’ but GNOME imo is far away from being taken serious. I know my words may sound harsh now but I see the project as a whole not the per person/developer basis. Many developer do a nice job on stuff such as Gedit a perfect esthetical application, MrProject (same), Gnumeric (once it solves the esthetical GUI issues as written in my review) and Evolution but there are also a lot of applicationjunk without any usage bundeled with GNOME there are far more powerful counterparts in KDE these days.
People in the GNOME community should seriously better organize (as it starts now – when it’s too late) otherwise GNOME will always suffer in the same ‘hackers toy’ cliche. They also need a serious release coordinator and spokesmen for the public and no insane individuals who force their view of things on other people.
greetings,
oGALAXYo
Cluttered doesn’t mean too much options. I personally believe that KControl is cluttered. Not because it has too much stuff, but it is just so much at your face – imagine how intimidating it would be for new users. And the names for some of the applets are quite confusing to new users, e.g. “Style” and “Themes” – wow, as a new users I can tell the difference… NOT. I don’t think that KDE should reduce configurability (I would move elsewhere otherwise), but rearrange things.
And KControl isn’t the only thing that cluttered. The menus too are cluttered. Not that it has too much options, it has too much redundant ones, and is poorly arranged. In addition to that, contextual menus have way too much in it. Personally, I think context menus should have options for that very thing you are selecting and right clicking (e.g. if I right click on a table in KOffice, I should get a context menu about tables).
Konqueror too isn’t free from my critism – it is too cluttered. Not menu options, just the amount of menus. Plus, when web browsing, a lot of the stuff on the toolbar and in the menus aren’t revelant, and the same way around with file management and file previewing. Plus, the menu bars have so many menus that barely have 3-6 menu items – what for?
I don’t mind integration, but not like this.
I can go on and on about areas that is cluttered in KDE – it isn’t because there is too much options, but because it is very poorly arranged.
oGALAXYo,
there are probably 5 times as many paid programmers and UI designers working on the GNOME project as working on the KDE project. GNOME is getting some serious financial backing nowadays, and has garnered more commercial support in the form of companies which use GNOME as their base. KDE is now beginning to get more finanical support and more companies are beginning to grow up around KDE, additionally various government agenices in euorope are also funding, indirectly, the implementation of certain features which governments need in order to make use of free software-much like Sun has helped GNOME in meeting the accessiblity requirements which the US government imposes on software contracts.
Please don’t get so frustrated- your comments and efforts were not in vain. But you must see one of the most important distinctions between free software and commercial software: Commercial software operates according to final product release schedules; software releases are held to be final, and only patches and fixes are generally allowed after release; anything beyond this is a new product, a product which earn money based up on its superset of functionality. In Free Software there is no such thing as a “final” release, unless of course the project is abandoned. The call for perfectionism, which is thoroughly justified when I have to pay $200 for this new final version is inappropriate in the context of gradually improving free software projects. Think of it this way- in the world of commerical application development progammers are trully happy when they have finished the product, whereas in the opensource/free software world no one wants a project to come to an end. Ironically open source/free software is a work model more closely akin to housekeeping than commercial software production- a housekeepers works is never done-if a house appears perfectly kept(ie. clean) it is because the housekeepers constantly clean,except of course when you are looking……….If you do not grasp this fundamental distinction between commercial and opensource/free software development-ie. between linear production and heuristic production- and you approach free software projects with the same expectations with which you would approach commerical software corporations-you do them and yourself a disservice- is not the fact that these two different forms of software development have different names not enough for you to recognize and respect the difference between them ?
In your first few sentences – I can’t really pull a representative quote from it as you have very long sentences – what are you trying to bring across? I am not really sure what your point is.
To want “experts” in cutting edge fields, where the the pace of technological change is rapid and where the markets are constantly re-adapting and restructuring is simply absurd. THERE ARE NO EXPERTS IN THE FIELD OF UI
Wow – you better tell that the ACM so they can get rid of SIGCHI. And all the product designers in the world.
Of course there are experts. HCI is not something new, people are thinking about how to communicate with computers ever since computers exist.
If you claim that there are no experts in HCI because it is a “cutting edge” field, then I suppose there are no experts in computer science either.
Most of what I do with my computer today was unthinkable 10 years ago-no one on the planet earth was capable of doing it- and what I do is so “normal” nowadays that its not even worth mentioning-you know like burning CD’s, making MP3’s, watching DVD’s, surfing the internet. When so much can change inside of 10 years, effectively throwing that which was common sense 10 years ago out the window, how can one pretend to be so “on top” of things today to believe that what they are spouting off today is going to have much validity in 10 years from now.
That is a very weak argument, considering that the UI you are using to do all those new things was designed about 20 years ago.
In the linux world there is little interest in “representative” user feedback. Linux is grassroots. Grassroots democracy means bringing the decision making to the people-not taking it away from them.
Democaracy is all about voting representatives who are supposed to make decisions. That’s what parliaments and senats and governors are for.
Everyone making his own decisions is called anarchy.
This would also mean mean that I would have to refuse submitted patches and improvements from potential contributors until 5% of the total userbase inisted on the same changes-well there goes 99.9% of my potential contributors….
No. Representative feedback is feedback that tells you what most users want, not what the most vocal users want.
The MS Windows metaphor cannot be carried over 1:1 to the linux world-because beneath the GUI in MS lies *nothing* and beneath the GUI in linux lies *power*. The same hold for MacOS prior to OS X.
Could you elaborate a little more, especially on that MacOS part? How is the UI affected by wether a Unix kernel or a NT kernel handles scheduling and memory management? The whole point of UIs is abstracting things so that the user doesn’t have to worry about all the technicalities.
Hi,
Well the times where I used to be totally frustrated are luckely over. I now accept the things as they are since I as user can’t change them. I know that the GNOME projects is backed by SUN with the A11Y (Accessibility) stuff and that various SUN employees are also working on the UI stuff such as the HIG. The problem is that we should NOT play down the fact that there are some minor design wrong decissions where made which affects a lot of applications these days. Since I can’t explain in detail because of my bad english I gonna try to give you some easy examples that reflect best what I’m trying to say:
a) Professionals from SUN are working on the HIG which is definately a need and must in the GNOME community for development. Side-effects are that various aspects written inside it are cut out of nowhere which leads into heavy discussion between the KDE and GNOME people these days. One argumentation is the button ordering. 2 Leading desktops that work on a corporate Usability GUIDE who can’t agree on this point. We have a big problem here. Even with respect to the SUN employees working on these decisions but we must agree here that even these people regardless how much of an UI experts they are – are only cooking wit water here. Many of the decisions they made in the new HIG are based on the Macintosh Usability Guides. Basically they went away from the Windows feel and moved over to the MAC feel. There is no single outside user being involved in these decisions and no one outside can verify that these decisions are actually good decisions. The conversations about pro’s and con’s are comming up NOW, now that both projects are working together.
b) As I previously said the OO design of the filemanager was brought up again. Even after things got settled that it should follow the old hirarchical filesystem structure. Now the discussion comes up again.
c) Stuff as esthetics in the applications. Now you said that GNOME has a UI team 5 times as much as KDE has unfortunately the KDE people seem to do a better job here since esthetics are still not cut in the GNOME community. I’m sure that the UI people are well aware of the critical situation in GNOME and now after my review many more people found and eye for it but still again the discussion came up while GNOME is in it’s progress state.
d) The UI experts found some agreement how an easy interface has to look like and what should be put into the Control Center, in the Applications preferences and which should go into GConf. But that’s from the sight of the experts how does it look for the nonprofessionals who contribute their apps in the means of GNOME ? say not the applications in the core GNOME bundle, say the apps outside of it ? it’s hard to track all the issues.
e) Even now a lot of decisions where made on the GNOME mailinglists, decisions that could easily be put in the category of the beginnings of a planning phase not in the itterative process as we have it now. Say as example the Toolbar issue. This could easily have been nailed down months ago but it was realized now, 5 ways of making Toolbars in GNOME it may be easy to track the issue down in the core GNOME applications but look at all the 3rd party apps that inherit the bad code. I was told that some developers started to write wrapper functions but is this the best solution on the long run ? stuffing the libraries full with wrapper code ?
f) Simplification is indeed a must as described many times, the GNOME control center is cleaned up, only values to change for the people that are really a must and the user shouldn’t be confused with the overhelming amount of options. But did it really simplyfied ? I doub’t so. Who decides which preferences the User should change and which not, so who can confirm that the values choosen in the Control Center are values that make sense ? It will come the day that the user likes to change some indeep parameters and here he hits to another level of complexity GConf. A lot of keypairs in a Tree of values that reminds one of Windows Registry come up, values that are partially described and partially not. GNOME right now comes only with a handful of applications so all the stuff is quite in good position so the user can control it but measure it with the amount of applications that KDE delivers for example. As rajan r wrote previously the KDE Control Center is overhelmed with stuff but that’s because every application adds its capplet there. But the enduser only needs to check for all his preferences in one location. The control center only and he can easily say that this option belongs to this capplet and not to something else.
Ok these are only a minority of things that I took out of my mind when writing it down. For sure not everything was viewed indepth or explained the precise way I liked to do it but only some points to think about it. Having 5 times as much UI people doesn’t mean that the decisions are good or that it makes into the code. There are a lot of people working on GNOME projects and contribute to GNOME who doesn’t necessarily like everything in the HIG or the idea of simplification. Convincing those are a problem. In a POS world where the customer pay for a project such as MacOS-X or Microsoft Windows there is a different view of things. Bugs may not get fixed as quickly as in open source but their developers are getting paid for their work. They spent all day in the project and at the end some quality crew containing 1000 of people will check the product and all issues will be nailed down. This is differently on open source imo. Back to open source and GNOME even these 5 Times of UI people are just people, they need that the developer cooperate with them. If they don’t then the UI people can’t do much either. Example: Totem, let’s say that the UI people suggest it to have less items in the panel but the maintainer and developer simply says NO, what can you do about it ? nothing. Thats the problem you first need to convince the people and have them understand that they work on a bigger task, not the hackers toy they played with for many years. They work on a complex system together with other people reaching one target. If you don’t like it as maintainer then it doesn’t mean that the rest of the world doesn’t like it either.
Gnome2’s non-configurability, even for correcting glitches, and KDE’s bloat and CPU usage, together with its maddenning slowness, have driven me to XFce, Windowmaker and IceWM. Blackbox is quite nice also. What do these have in common? They are each 10 times faster and more efficient than the ever-glitchy gnome and the topheavy KDE.
Before being flamed by the KDE tifosi
let me do some comments on “too much configurability”:
I’ve got myself one of those Knoppix-based CD live distros. I booted it and started working.
I can’t stand the single-click behaviour, so I tried to turn if off… Man, I had to navigate sooooooooooo much on the KDE Control Center! Too much options with the same name. Toooooooooooooo bad.
Both are right about certain things….
Eugenia is correct concerning KDE. But KDE’s bloat and CPU usage is my tic. I simply dumped it, as I dumped Gnome and its inherent glitchiness and useability problems.
Windowmaker is not only faster, but more configurable also, which proves that simplicity, in code design, reigns.
XFce, with its X-based apps and configuration tools that reach the core of X11/XFree, has even them beat for system-wide configurability.
If I smoked more, I would use KDE which would give me time to roll and smoke a cigarette during the launching of apps.
…people re-reading http://mpt.phrasewise.com/2002/04/13 and http://mpt.phrasewise.com/discuss/msgReader$182 on ALL that discussion.
Please investigate and learn the difference between a Windowmanager and a Desktop Environment. These mistakes are being made over and over again. People compare apples with pears and that’s not right. Kwin and Metacity as standalone windowmanagers would probably be was fast as IceWM or Blackbox and if you use IceWM or Blackbox inside of GNOME or KDE then things won’t be faster compared the environment as a whole. The speed issues for the applications is also not in responsibility of either GNOME or KDE they at the final end fall back to Binutils (Linker, Assembler etc.) and GCC (C and C++ Compiler). Some parts in the Linker and Compiler are specially heavily improved for C applications and other parts such as C++ is badly improved so the resulting code is shitty slow. At the final end I doubt that even IceWM and Blackbox are so much of an advancement specially if it comes into mixed application usage such as Mixing KDE and GNOME apps. At the bottom layer KDE and GNOME apps depend on large chunks of various libraries that get loaded on demand. I with using GNOME only have the advantage that all my applications (I use GNOME only here) using and accessing the same libraries which makes eating of resources at the end acceptable (still to much). Same goes if you use KDE only. But as soon as you start mixing applications the way you do on your fast IceWM (while IceWM or Blackbox are only freaking Windowmanagers who are NOT significantly responsible for applications loading up fast or not) you will waste more resources. Start Evolution under Blackbox and then Konqueror and you waste more resources in this second than people who use KDE or GNOME only. Specially people who say that by using Blackbox and stuff like Openoffice (loads its own different libraries), Evolution (which loads half of GNOME libraries) and Konqueror (which loads half of KDE libraries) are faster than their counterparts in environment only such as using KDE or GNOME only need to sit down and do better homework because they basically have understand nothing.
> But KDE’s bloat and CPU usage is my tic.
This is also a wrong statement. Neither my GNOME nor my KDE are trashing CPU usage here on my system. Both in idle state (when loaded up and ready for user interaction) idle around 1% in it’s worst case. for an XP1600 this is around 16Mhz wasted CPU power. When it comes into user interaction e.g. Loading up Applications then this is only an issue of:
– Binutils (Linker, Assembler)
– GCC (C++)
for not generating highly optimized code from either C or C++ source code
– Harddrives
for people using shitty 7.2 GB harddrives that only runs 2 mb/s (or 500 kb/s if they are too stupid to turn on 32bit mode with hdparm and DMA).
– Slow machines
for people that tend to use trash for serious work
– Less Ram
times where 32mb was enough are seriously over.
And please again. If you use IceWM then starting of Evolution will take the same 10 seconds under IceWM than under GNOME itself (while I belive that starting the same program under GNOME will be faster due to preloaded libraries). Same goes for Konqueror while it takes you 10 seconds to load up under IceWM it will take the same 10 seconds under KDE (whicle I also belive that it will start faster because of preloaded libraries).
To proof that the Windowmanager IS NOT responsible in speeding up applications or faster loading you can substitute IceWM with any Windowmanager of your choice and you measure the same loading and execution times. A proof that the Windowmanager is only responsible for what it does. Managing shitty windows and nothing else. This is a valid statement if you have enough ram and a fast harddisk. If you use 32mb of ram and use different Windowmanagers that eat itself a lot of ram then the OS starts to SWAP ram to harddisk which slows down again but in an ideal world (as many of us already have) with bigger machines and good ram this is not the case.
Quote
“If you claim that there are no experts in HCI because it is a “cutting edge” field, then I suppose there are no experts in computer science either.”
well let’s be a bit more specific shall we. Computer “science” has existed how long…hmmm….if you tie computer science to the invention of binary logic we can go all the way back to Pascal(18th century)…but more likely you mean the field of cybernetics which started in the 1950’s. Ok so lets say we have 50 years of computer science, ok? now how long has a sizable population been using a graphical user interface-well the Palo Alto Xerox guys were doing this in the late seventies but it wasn’t until the early nineties that most people associated a GUI as their primary interaction with computers. So lets say we have 13 years or genuine GUI experience-and even if I were to date this back to the Lisa, the first commercially sold consumer computer which came with a built-in GUI -was 1982, and that gives us only 21 years……
Quote
“That is a very weak argument, considering that the UI you are using to do all those new things was designed about 20 years ago.”
No its not, it is a very, very strong argument-the entire nature of human-computer interaction has evolved fundamentally over the course of the last 20 years- someone who was a great programmer o n IBM mainframes back in the days of punch cards has little if any utilizable knowledge in todays computer world-unless they retrained themselves along the way. No industry has so rapidly discarded its immediate past into oblivon with successive waves of changes rendering previous knowledge obsolete-as has the computer industry-nothing even comparable.
Quote,
“Everyone making his own decisions is called anarchy.”
non-sequitor. The issue here is not wether or not each person can or cannot make their own decisions. This is a fact of life-each person makes his own decisions. And yes Linux is more akin to anarchy, in the sense of spontaneous order, than it is too a regulated beuracracy, like corporations. What people fear most of all about the dreaded word Anarchy is the notion of a Hobsian-inspired war of all against all. This is not what is happening in the linux world at all. Yet it is somewhat representative of the capitalistic venture-capital based software industry which constantly resorts to war like tactics to kill off their competitors….
Quote,
“No. Representative feedback is feedback that tells you what most users want, not what the most vocal users want.”
I trust self-selection at least as it pertains to open source/free software projects. Such is inappropriate for commerical applications-they are paid to be indifferent to the particular wishes of particular users….
Quote,
“Could you elaborate a little more, especially on that MacOS part? How is the UI affected by wether a Unix kernel or a NT kernel handles scheduling and memory management? The whole point of UIs is abstracting things so that the user doesn’t have to worry about all the technicalities.”
In the linux world the point of the GUI is not abstracting things so that the user doesn’t have to worry about all the technicalities. Certain things can only be done at the CLI level, other things can only be done at the GUI level, there is a large proportion of overlapping between the two, in some OS’s where there really is no CLI (ie. windows XP and MacOS prior to MacOS X) -in which case that which cannot be done at the GUI level simply cannot be done. Linux is usable without anykind of GUI, in fact many, many systems make no use of GUI at all. UI is a very broad and general category under which the GUI and CLI are mere subcategories-as such it remains abstract and negates the differences prior to their specificity-the move from the CLI which all computers used 20 years ago to the dominance of the GUI today has been the re-definition of what computer “use” means and what “users” are…….
So lets say we have 13 years or genuine GUI experience-and even if I were to date this back to the Lisa, the first commercially sold consumer computer which came with a built-in GUI -was 1982, and that gives us only 21 years……
What does that point have to do with commercial success or age? Your simple claim was that there are no UI experts because UIs for computers are too young? So – we have no Java experts, no experts on web servers? HCI is a well-researched science and the way humans think and perceive doesn’t change very quickly. HCI is not about computers – it’s about humans.
In the linux world the point of the GUI is not abstracting things so that the user doesn’t have to worry about all the technicalities.
Sure it is. Any UI, graphical or not, is doing that. The bash shell does lots of abstractions already – it abstracts the underlying kernel and libraries.
in some OS’s where there really is no CLI (ie. windows XP and MacOS prior to MacOS X)
Windows XP has a CLI. And the CLI is not necessarily part of Unix systems – you could very well build a system based on a Unix kernel without a CLI if you wanted to.
in which case that which cannot be done at the GUI level simply cannot be done.
An example of what cannot be done? After all, both graphical and text-based UIs are using the same fundamental APIs, where’s the limitation?
W000t, yet another Mosfet article.. The typical graphics designer who thinks he can make a good GUI. If you make it flashy enough, it will be user friendly.. Set font size to 8, that looks prettier! (ignoring the fact that only people with perfect vision and monitors will be able to easily read it)
Users should be able to configure every single aspect of GUI look and feel! To me that sounds like making the users into a sort of developers.. Users *use* the program, developers *develop* the program. If a program doesn’t work or feel right, users should contact the developer. They shouldn’t have to fix a program themselves, it should just work.
Do you have people who prefer the Windows way? Let them use windows! Instead of giving your desktop environment a multiple personality disorder..
> Users *use* the program, developers *develop* the program.
.. and how many times did developers develop programs away from user needs and make it unusable for users ? This is happening permanently on open source ..
Yeah I speak Mosfet a higer value of knowing what to do than Havoc Pennington. But this doesn’t mean that I go personal or something it’s just an opinion that I share with Mosfet which I think sounds more reasonable than what 2nd said does. We the GNOME users are forced to use a lot of junk-inventions made by 2nd named person.
From http://developer.kde.org/documentation/design/ui/summary.html
Avoid rampant customisation.
Customisation has the effect of delegating part of the interface design to the user. They may or may not be qualified to do this. Users are not usability experts; we are. If a user can, by a few judicious choices, really improve the interface, we probably have done a poor job. Most users, if given a reasonable interface, just want to get their jobs done.
I wish they’d stick to that.
. and how many times did developers develop programs away from user needs and make it unusable for users ? This is happening permanently on open source ..
If only you read exactly one sentence further. Really, just learn how to read.
Quote,
“So – we have no Java experts, no experts on web servers? HCI is a well-researched science and the way humans think and perceive doesn’t change very quickly. HCI is not about computers – it’s about humans. ”
1) Java is a computer language – all computer languages can be learned and mastered and yes there are probably many, many java experts; In a field which is clearly delimitable-ie. the scope of a computer language, one can reasonably speak of experts in that field.
2) Webs servers are concrete examples of computer application, speaking of experts as regards web servers is extremely appropriate, even more so than speaking of a java expert- for that which is entailed in being someone who administers or designs web servers is even more specific than someone who writes java code-which can be used for 10,000’s of different uses-of which writing web servers is just one.
3) No doubt, much research has been done in HCI. Yet one overlooks one simple thing concerning the complicity of cognitive “science” and cybernetics: Cybernetics was born during 1950’s; the terminology coined during that time to describe the functionality of computers, and more importantly what functionality computers need to implement, was based on our then current understanding(ie. theory) of how the human mind “functions”; Fifty years later we now use computers “analogies” to explain how the human mind functions; As if there was any relationship between that which memory means for human beings and what RAM means in terms of computers. This is a wonderful example of two logical erorrs: 1) post hoc ergo propter hoc and 2) tautologie. Just because the words we use to describe the functioning of computers are the same words which we now use to describe the function of the human mind does not mean that a) we understand how the human mind “functions” or b) that we fully understand the nature of human-computer interaction.
Quote
“Sure it is. Any UI, graphical or not, is doing that. The bash shell does lots of abstractions already – it abstracts the underlying kernel and libraries. ”
we are confusing semantic levels here. You speak of UI in general, I have been speaking about GUI vs. CLI. In a generic sense a UI is an abstraction of system level components which render these components accessible and usefull without having to deal with the system as such. I am fully aware of what you mean here. But for the record where does, according to what you have learned, UI start and at what point does it stop ? does the UI only apply to applications ? or to programming languages ? or compiled binaries ? or at the microprocessor ? or at the level of design of the motherboard ? ie. where does the User Interface with the computer-at the keyboard ? via the mouse ? via the screen ? Don’t assume I am ignorant of how UI has been defined.
Quote
“Windows XP has a CLI. And the CLI is not necessarily part of Unix systems – you could very well build a system based on a Unix kernel without a CLI if you wanted to. ”
well there is a sorry excuse for a CLI in WinXP, but I am reluctant to even call it such, due to its profound limitations. Yes of course-microwave ovens can have an embedded microprocessor running linux without CLI.
Quote,
“An example of what cannot be done? After all, both graphical and text-based UIs are using the same fundamental APIs, where’s the limitation?
The limitation is everywhere where the GUI is better suited for certain tasks than the CLI-and vice-versa. Hardly anyone does graphic art using the CLI-at least nowadays. Hardly anyone does hexadecimal programming using the GUI-unless of course they are using a GUI representation of a CLI……
the move from abstract to concrete and from concrete to abstract is not a 1:1 relationship; There exists differences between ideas/theories and the world in which we find ourselves in…..even when we attempt to negate these differences through the use of common significations.
KDE is too bloated & half broken for me to bother with, i do use Dropline Gnome for Slackware-8.1 but without Nautilus so for a file manager all i got is mc in a terminal (fine with me)…
i now keep an eye on WindowMaker & ICEwm as decent and light WMs for Linux and use them both frequently…
the last really good release of KDE is 2.1.1 that is included with Slackware-8.0 and Redhat-7.1 & Mandrake-7.2
Gnome-1.4 should be continued to be developed to keep it light…
I gotta agree with Havoc on the options issue. As a software developer it is my experience that can and do increase code complexity and result in a lot of errors.
Here are 2 fundamental rules I know about UI design.
1. Designers are not users (User input is important)
2. Users are not designers (don’t blindly follow user input)
While Mosfet seems to understand #1, he completely misses on #2. Traditionally, Unix/Linux programs have suffered from the designer not understanding #1, but I think things are now swinging in the other direction. Good UI design involves making hard decisions about what features to leave out.
My take on the comparison article is that most of the issues about Gnome 2’s consistency are due to Gnome 2’s immaturity and the more distributed nature of its development. Gnome 2 is a platform, and a new one at that. KDE is a complete system. With KDE you get a desktop and a full set of apps (KOffice). The desktop and most of the apps come from the same place, so it is no wonder that they are pretty darn consistent. This is made worse by the fact that there are currently few real Gnome 2 apps and the configuration utilities are far from finished. The Gnome 2 release was really more about the underlying pieces than the whole desktop. I expect this to improve in the 2.2, 2.4 and 2.6 releases.
All of this adds up to Gnome 2 currently running well behind KDE in the usability department. However, I think KDE may run into chaos problems due to all of the options and Gnome 2 may begin to gel into a more useable desktop (once the apps and config tools catch up).
…between window managers AND desktop environments.
A window manager will ALWAYS be lighter than a desktop environment because won’t carry so many services to build a consistant desktop.
To be REALLY fair:
1) You should compare KDE and GNOME with XFCE and maybe GNUstep and Enlightenment DR17 (I don’t know the current GNUstep state, and DR17 is on CVS for a long time)
2) You should compare Metacity (GNOME’s WM) and KWM (KDE’s WM) with WindowMaker, icewm, Sawfish etc etc etc
I do know the difference between a desktop environment and a window manager. I use XFwm with just about anything. Another fact is that KWM is by far the worst WM out there. True: once the agglomeration of apps called KDE is loaded and ready to run, I only have time to roll the cigarette and light it, but not to smoke it. The problem with KDE is piggish bloat, crashy libraries, a bad wm and its ressemblance to winblows. Gnome is much faster. I can launch any Gnome app from Wmaker, IceWM, Blackbox, or XFce as if Gnome were running. KDE is fine for folks who like Windows, have a lot of time to waste, and want hundreds of worthless applications that are only compatible with themselves.
I have used Gnome since the inception of Gnome…, same with XFce and Wmaker. But since Gnome2, I dumped Gnome, though it does run quite well with any Gnome compatible WM. I do not use Gnome, but I have all the pertinent libraries. Some of the apps, especially GVim, Galeon, etc….
I I found that KDE runs very well on PC’s made for Windows’ bloat. As far as CPU usage goes, this happens when a KDE app is crashing or “semi-crashed”, wich is quite often.
In spite of what you say or mistakenly believe, the combo XFwm + XFce is far superior to Gnome for a serious programmer, that is. WindowMaker has them all beat in ease of configuration and speed. You are badly mistaken when you say that KDE is “fast”…. or you are living in an ideal fantasy with your eyes closed.
So to sum it up:::
KDE suX, SLOW, but it’s pretty, which makes it appealing to children…
Gnome is great, but very very very very very glitchy…
XFce is great
Wmaker is great
Blackbox is very good
IceWM is very Good
KDE SuuuucKKKKKKKKXXXXXXXXXXXXsssssssss
KDE dont suck, its only you who think it does… If you want to use a window manager instead of a DE, just go ahead and do it, noone is going to stop you. It doesnt make KDE crappy just because you think so.
That Simpsons episode ( http://www.snpp.com/episodes/7F16.html )where Homer meets his long lost brother the car manufacturer? His brother (Herb) let Homer design a car for the everyday joe. The resulting car with all the bells and whistles and contraptions cost too much money a drove Herb out of business. While I know it doesn’t quite match up to Mosfet’s ideas, it’s still a pretty funny comparison.
To stay sort of on topic. As a power user, I think I’ve got to side w/ Havoc on this one. He makes a much stronger argument I think. GNOME 2.2 is on the right track and as soon as the apps catch up, I think it’s going to be the platform to beat in the future.
I think some KDE options could be more consolidated, since people would throw a fit if they were trashed. Styles and themes, for example. It just feels like Kontrol is scattered, and puts related things far from each other. On my K6-2 350 MHz, going through the menus once to change all my preferences takes too long. At no point do I have a couple of hours to configure my desktop from the “broken default” just because the Konfiguration program is whacked out.
Just because the words we use to describe the functioning of computers are the same words which we now use to describe the function of the human mind does not mean that a) we understand how the human mind “functions” or b) that we fully understand the nature of human-computer interaction.
Of course, just like psychology or physics, HCI is a discipline that uses models and assumptions to describe things. Even though we do not know the “real thing”, these models work quite well for our needs. Human minds most probably do not have three idenities ich, über-ich and es, but it is a useful model. The same way, HCI has it’s assumptions and models that cannot be proven like math, but confirmed in studies. Just because HCI is not a 100% correct discipline like math doesn’t mean that we can ignore the research done so far. There is quite an amount of knowledge and experience that has been collected and we have people that are more intimate with that knowledge than others. In my opinion, it’s therefore legit to call indiduals that excel in that discipline “experts”. (although some tend to refer them as “gurus”)
But for the record where does, according to what you have learned, UI start and at what point does it stop ?
It starts where the user ends. The keyboard is a hardware part of the UI, so is the mouse and so is the screen. If you’re using a microwave oven, the buttons and maybe a LCD are the UI. Where does it end? That depends on the engineer…I have seen bad mixtures of UI and backend code
well there is a sorry excuse for a CLI in WinXP, but I am reluctant to even call it such, due to its profound limitations.
Just because hardly anyone uses it doesn’t mean it’s crap. And if XP’s standard command line doesn’t suit your needs – install Cygwin. Bash and other shells are just a program, just because it doesn’t ship with the system doesn’t mean it doesn’t exist (MacOS 10.0 required post-installation of bash too).
oGALAXYo:
KDE has made some quite bad decisons.
Firstly, it is questionable whether a file manager should be doubling as a web browser. The only other browser which does that is Internet Explorer, which is now more or less the same thing as the File Explorer in Windows.
Secondly, you get Gconf wrong. I actualy think it is a good idea to have a system store it settings in one place. GConf should never be compared with the Windows registry. GConf does not make use of unintilligible keys. The are quite ‘English’ and very understandable.
Thirdly, I think it is bad for a desktop to ship with too many apps. The yare usually rushed and of beta quality. Gnome must be commended for not bringing out many of its apps already ported to Gnome2 for quality concerns. Apps should be available as addons. Basically, what is needed is a browser, file manager and maybe a media player. Apps should sit on top of the DE, and not have it all bundled there.
Gnome is producing some really useful projects, such as Gstreamer. They are more original than KDE. It is easy to copy. The KDE experience tries and fails to mimic the Windows/Mac feel. Gnome is quite its own direction and is a pointer to how original open source can be. I feel quite a lot of Gnome development is useful to the Linux Desktop in general, whatever the DE. However, all KDE development seems to be KDE-centric. For example, GStreamer, Fontilus.
It’s not only kde or gnome’s fault, lets not forget xfree, this is 2003 and there’s still no colored mouse cursors, there’s still no support for 32bit modes (alpha support), on the fly resolution and monitor frequency changing, all of these are still in beta.
Not to mention the bad responsiveness, horrible redraws, bad support for binary drivers and no easy way to install them etc.
Gnome of course works with directfb, a far better solution for the desktop but it’s plagued with the lack of drivers.
Well you are getting the points wrong here. There are definately a lot of nice things being invented by GNOME but most of the stuff never comes out of the development phase. Once it get’s shape and usable the developer start over from scratch and change everything in that app, which makes it unusable for productive day. Personally I like the idea of having the Filemanager and Browser in one application – switching between Tabs and use one Tab as Filemanager and the other to Browse the web. The idea and implementation is absolutely cool but it’s for sure of personal taste.
Secondly, no, I don’t think that I get GConf wrong. This Gconf issue is a exasperating topic. It has been invented and with some ideas in the back but it has never been proven to be the optimal solution. At the final end it’s currently used as the old gnome-config way for the single homeuser. The argumentation to have it used by administrators to set global values is cut out of nowhere and is only a theoretical aspect that gets discussed by people but it has practically never been proven to be used that way. Even the fact to be able to use different backends is also more or less a theoretic aspect and luckely no-one wrote a different backend for now. The day this happens, the day other backends get used is the day where you simply can’t tar your gconf dir up and move the contents to your bureau for business work. At the end there are many more issues with GConf than with the old way of doing configurations. It’s questionable if the new features like setting keys globally also on other preferences is such an good idea. As a programmer view I’m able to come up with a lot of disadvantages by using GConf but I also know by presenting all the known negative aspects of it, that it may sound like a rant or troll in the eyes of pro-Gnome readers of this site. People easily tend to value true facts as troll or rant after they realize that the brought up points are with value. So it’s easier for them to argue in that way than dealing with the critics. You are welcome to read what KDE developers think about it.
Last, you said that KDE experience tries and fails to mimic of Windows/Mac feel and I must answer that you are totally wrong. I think that KDE immitates Windows perfectly and if not better in certain areas. There is nothing wrong by doing this. Windows is used by many people all over the globe with success, they do their business work and they seem to be happy with it. The whining and prejudices by people that fail to maintain and install windows are not worth the reaction since these people have no technical skills to decide or talk about computers at all. It’s shortsighted and wrong to ignore the inventations that Windows made. Many of their developers are actually as good Linux and Unix experts as many of us. These people practically got all their skills from such machines. GNOME is more and more following the MacOS-X way. I know and am able to prove this fact since many developers who work on the heart of GNOME are using MacOS-X as their primary operating system on their lappy or their desktop. You may realize why GNOME has to look the same way.
> lets not forget xfree, this is 2003 and there’s still no
> colored mouse cursors, there’s still no support for 32bit
> modes (alpha support), on the fly resolution and monitor
> frequency changing, all of these are still in beta.
From information side of view. Xfree as is in CVS does support colored mouse cursors.
The 32bit issue is NO issue since the real depts is indeed 24bit the Hardware of your graphics card usually uses 32bit because its faster to calculate for the processor to sent the data there. 32bit is one longword and the processor only knows about byte values (8bit), word values (16 bit) and longword values (32bit) to sent 24bit you need to sent a word 16bit and then a byte (8bit) after this which takes more calculation time than sending one longword. At least that’s what I still recall from the past but I am not 100% sure if I’m right here but finally does it make any change if you use 24bit true color on Xfree or force your hardware to use 32bit data while it only shows 24bit ? I would be thankful to get enlightened here.
> Not to mention the bad responsiveness, horrible redraws,
> bad support for binary drivers and no easy way to install
> them etc.
Why ? You are dealing with Linux/BSD and other Unixes here why do you want easy install of drivers ? Grab the Kernel sourcetree and compile the kernel of your needs the stuff you like and want will probably never happen on Linux/BSD and the other Unixes because no one besides you and the JOE user wants it. Many people are missguided here they by mistake assume that Linux is the next Windows which is not. It’s a totally different Operating System architecture which has it’s roots back in the 60’ies which learned a big revival in the early 90’ies. That’s what makes Linux so attractive. If you really don’t like to spent the time into configuring a Linux system optimally on your demands then you have chosen the totally wrong system.
> Gnome of course works with directfb, a far better solution
> for the desktop but it’s plagued with the lack of drivers.
Again what drivers ? GNOME is no Operating System it’s an Desktop mounted on top of existing Operating Systems. Your way of argumentation leads into the assumption that you are a bit confused about how your system works. Some good books explaining the basics of Linux may help understanding it better.
>I would be thankful to get enlightened here.
It’s 24bits for RGB and the extra 8 bit are used for transparency (a.k.a. alpha)
>You are dealing with Linux/BSD and other Unixes here why
> do you want easy install of drivers ?
Why the hell not!
Again what drivers ? GNOME is no Operating System
I meant directfb drivers, not gnome drivers.
Do you actually know Konqueror?
> Firstly, it is questionable whether a file manager should be doubling as a web browser. The only other browser which does that is Internet Explorer, which is now more or less the same thing as the File Explorer in Windows.
You might be horrified to hear that Konqueror itself is only a useless program shell which uses a kind of plugin system called KPart to show and allow users to access a wide range of protocols and systems. There are KParts for the file system, many archive file formats, protocols like http, ftp, pop3, imap, telnet, smb, nsf etc., funky wrappers like help, man, audiocd, camera, perldoc etc. KPart is available to any KDE app as well as all open/save dialogs with KDE which means you can remotelly open/save files of any supported protocol etc. The possibilities are unlimited while the core, Konqueror itself, is very light.
This highly flexible system is the reason why khtml, used for html redering, contains no bloath but can refer to other KParts for additional services.
A list of all KParts on one’s computer including additional infos can be showed using the command “kcmshell ioslaveinfo” (start a terminal or press Alt+F2 in KDE).
You might be horrified to hear that Konqueror itself is only a useless program shell which uses a kind of plugin system called KPart to show and allow users to access a wide range of protocols and systems. There are KParts for the file system, many archive file formats, protocols like http, ftp, pop3, imap, telnet, smb, nsf etc., funky wrappers like help, man, audiocd, camera, perldoc etc. KPart is available to any KDE app as well as all open/save dialogs with KDE which means you can remotelly open/save files of any supported protocol etc. The possibilities are unlimited while the core, Konqueror itself, is very light.
That’s all fantastic, but irrelevant. We are talking about the GUI here, and usability. Doesn’t matter to the user how it has been implemented.
Yes, it is irrelevant indeed since it was an answer to an already off topic post. Thanks for adding yet another irrelevant post. =P
thank you for your well reasoned response
Quote,
“There is quite an amount of knowledge and experience that has been collected and we have people that are more intimate with that knowledge than others. In my opinion, it’s therefore legit to call indiduals that excel in that discipline “experts”. (although some tend to refer them as “gurus”) ”
My main point of contention here would be that the open source/free software world of software has been slowly, but surely, re-writing much of the logic which has hitherto driven the computer industry and as such represents a break within traditional computer usage paradigms.
Many of todays opensource/free software applications and OS’s themselves arise through the more or less spontaneous happenstance of multiple people sharing the same idea at roughly the same time, geographically spread around the globe in dynamic online communities bereft of centralized structures capable of imposing top-down planning of software development.
If one learns in HCI that only such development can lead to good UI’s -which is an argument I here time and time again from UI experts- then HCI will be re-writing its theories again to remain relevant. “Gurus” are charismatic leaders capable of persuading their followers into believing that their way is the “right way”-which is what irks me when one talks about UI experts.
These experts may be accurate, ie. reflecting a high degree of correlation between their theories and their empirical observations. Yet because of their assumptions about how good UI software must be developed and what is representative of good UI they remain largely irrelevant to the actual development of software in the opensource/free software world where loose-knit structures tend to be formed around degree of participation and meritocracy as opposed to commerical software production which is charachterized by production-based division of labor, ie. heirarchies, within which context such things as UI experts first gain their legitimacy.
If such open source/free software projects do not entail a sufficient degree of division of labor to account for UI designers having a particular (vaulted) position within the process of software development is this a fault of the open source/free software model ? Or is it indicative that UI designers rightfully have their place in the context of commerical software development ?
The debian install scripts are certainly the problem this guy had with installation. I installed Gnome 2.2 on Gentoo just fine. All compiled from source, speedy, and working fine. However, I don’t currently use Gnome due to a screw up in nvidias newest driver that rapes your menus and the fact that it is slightly painful to go backward in nvidia drivers (I have no idea why).