KDE 3 was released only a few months ago, and it is, to date, the most successful version of the series, serving more than 50% of the Unix and Linux desktops, surpassing Gnome (~21%) and the rest of the gang. However, KDE is not perfect, and still not as comfortable as the Windows or as sexy as the MacOSX desktops. It lacks two things: integration with the underlying system and UI polishing. Today, I will mostly talk about the polishing part, as a lot has been already said elsewhere about the seemingly unsolvable integration issue (because of the modularity and completely independant/remote software projects.) Update: And as I was just publishing this article, KDE 3.1-Alpha was released. I hope that some of my recommendations will make it to the final version of KDE 3.1.
User Interface?
I served as a User Interface designer while I was working in a knowledge system project in UK and I worked as a web developer and web designer for the last 5-6 years. This article is merely a suggestion to the TrollTech and KDE developers on how to polish the current UI a bit. In some places, I might suggest quite significant changes, while in others, the changes would be trivial and as easy as pie, but still important. I hope that these developers will read my recommendations and we won’t fall prey to the popular “KDE developers do not care about the users” mentality that has headlined several news sites lately. This is just an article with some constructive critisism.
What is a good User Interface? Well, in order to answer that, we will have to take into account that different people like different shapes, colors and functionality. This article is just my personal opinion, how I would like KDE to evolve in the future. I am sure that other users would like to see other, different types of evolution. However, we can’t deny the fact that some basic rules of UI design should never be ignored. Aside of personal opinions and preferences, there are some basic UI rules. The problem is, KDE, by not being… perfect (who’s perfect anyway? :), does exhibit some problems in its UI.
This is a screenshot of my modified KDE 3.0.2 installation
Desktop Context Menu
Let’s start with the Desktop context menu dialog. In my opinion, this context menu is bloated with too many options in the root level, some unecessarily placed there, as they belong elsewhere. Example: The “Enable Desktop Menu” option, while it might be an important option for some (however, I never met anyone who actually uses that screen space sucker Desktop Menu), it is not an option that you will enable/disable all the time. It has no place in the main context menu. Its place is on the “Configure Desktop…” preference panel.
As for the “Configure Desktop…” itself, it should have been called “Desktop Properties” or “Desktop Settings”. “Configure” is a verb. Configure implies that the user knows how to “configure” something. Believe me, for a Unix newcomer, “configure” is a dreadful verb. It is a scary command. It might sound funny to you, but never underestimate the psychology of the user. UI is all about psychology. It is all about shapes, colors, pictures, words… Picking the right elements each time is the right way of creating a comfortable desktop environment.
Continuing with the context menu of the desktop, I would like to see the whole bunch of options from “Unclutter Windows” down to “Refresh Desktop” to go under a new submenu. Having these five options on the root of the menu, it just makes the whole panel too cluttered. These options are obviously related to Desktop actions, and they should be together under a new submenu.
What I would additionally like to have in the desktop context menu instead, are the following options:
1. Browse /home. As you can do with BeOS’ Tracker and the KDE’s “Quick Browser”, where by opening submenus, you can navigate to your ~/.
2. Browse /. Same as above, but allows you to navigate through the whole disk. This should only be available for the root user. In other words, put the “Quick Browser” on the desktop context menu.
3. Mount. Under its submenu, list the whole deal of partitions and devices that can be mounted and allow KDE to mount them on demand. Available
only for the root user or for the user with the right priviliages.
4. Add-Ons. This submenu, should be similar to Nautilus’ script folder and to BeOS’ Tracker Add-Ons. Some KPart add-ons (not necessarily stand-alone applications) will be able to do some great utility job. For example, select a few files and then pick these addons, addons like a GUI grep application, KRename, Burn to CD, Open Terminal (in current folder), Run Command (yes, this should be here and not on the root menu), email to, Compress, and whatever else you think it would fit as useful tool that can be applied to files and/or directories.
Additionally, I do not think that the “Help on the Desktop” is a good idea to be there at all times. It may be wise to leave the option there for the first week of using KDE, but then it should automatically removed from there, and only be accessible via the central Help docs.
The “Create New” submenu needs some polishing as well. Seperate the system entries from the entries created by third party applications with a menu seperator. For example, “Link to Application” is down the end of the menu, while it is something used quite regularly, and much more than the “CD/DVD-Rom Device” option which is more accessible from that menu, but it would only used pretty much, once every other moon.
Icon Context Menu
For the context menu when you are right clicking to desktop icon, I would suggest that as KDE’s got a special menu for the Trash icon, the same should be for any directory link on the desktop. For example, almost everyone’s got a “Home” folder link on the desktop. The deal would be to be able to “Browse” via the context menu in that directory, and wherever you click after that, to open a new window in Konqueror or Terminal on the directory you browsed to. Do the same for any other folder/device icon on the desktop. Allow to browse in it and then open it with Konqueror or Terminal if necessary.
Overall changes for the icons on the desktop:
1. “New Window” should be renamed to Open or Run.
2. Create a “Copy To” option that is able to navigate on your home folder and then copy the selected file(s) wherever you selected. Default navigation root for the user should be its Home directory, and / for the root user. Add the same option in the context menu for all files/directories/icons, on the Konqueror view as well, not just for the desktop.
3. Do the same as above, but this time for “Move to”.
4. “Edit File Type” should be part of the “Properties”, not stand alone. Properties panel needs almost complete redesign as well, but I won’t get into it this time.
5. “Open With…” should be a submenu and has listed the other optional applications (as found already in the system’s settings) that are able to open the selected file(s), and after using a seperator, then offer the option to open the panel to manually select which application you want to open your file. The way it is now done, it is too much time consumed for the user, while the whole functionality of this option and the alternative opening applications is already there, found on the “File Assosiations” Konqueror’s KPart!
6. The Add-Ons submenu with the useful utilities, should be available in this menu as well.
Kicker
Kicker is pretty powerful, but its default set-up is far from satisfying. There are, by default, 6-7 big icons on the left hand side, that auto-zoom in a pretty ugly way (zooming from 32 pixels to 48 pixels, without an intermediate “morphing”/transition that would help in the smoothness of the actual zooming). First thing I do after I get to a new KDE version, is turn that feature off. MacOSX has a similar zooming capability for its dock icons, but at least there, the zooming is done in a nice way, using vectors. Either try to get the zooming right, or do not put it there as default, please. Update: Wow, I just saw this!
My second problem with Kicker is the big buttons next to the “K” menu. They are huge, taking half the Kicker’s screen space on a low res monitor and make the K icon look like just another button, while it is an important menu and should have a seperate look to distinguish it from the rest of the big icons. I would like to suggest to move to the “Quick Launch” solution for these additional buttons (like the terminal, Show Desktop, Konqueror, Kmail etc) which in a medium sized Kicker (48 pixels), it can accomodate two rows of icons at 16 pixels each. There is already a Quick Launch feature on KDE 3 today, but it is buggy and it does not behave as it should (icons just shuffle around by themselves, you can’t add an application that is not part of the KMenu etc). Also, it is important to remember and understand that most Joe Users never ever modify their desktops (or even have the right to modify them in a corporate environment). 90% of users just stay with default settings. This is why the default settings have to be right and serve everyone.
The rest of Kicker is pretty well done, but I would like to suggest four more things:
1. The “Hide Panel” button should only be on the right hand side. Instead of having this huge 48pix (height) button that only hides Kicker on the right, “cut” the button in half and put an arrow looking on the left and one arrow looking to the right, underneath the first arrow. In essense, create two buttons there, which will only take the same screen space as before, but they will have the same functionality like when having two buttons.
2. I really like emulating the OSX Dock look and feel so most of the time, my Kicker is set to about 33% of space of my 1280×1024 desktop. “Expand to Fit Required Size” is ON, however, when I open a new application, Kicker does not resize to accomodate a 150 pixel widget of the Taskbar for the application that I just launched. This is a must-have feature. Kicker should be intelligent enough to know that Taskbar requires more space when a new application was launched creating an entry to the Taskbar, therefore it needs to resize itself. Instead, Taskbar today is just resizing itself (instead of just resize the whole Kicker), creating entries that are as small as 16pixels, making a centered Kicker feature, useless.
3. Please start grouping similar items after 3, 4 or 5 (user selectable) same applications are open, not just after 2.
4. The “tear-off” KMenu is ugly. This dotted option is really archaic looking. Please change it to something more soft, maybe something similar to the shot following down this page.
The Open/Save Dialog
I believe that the biggest problem in this dialog, and a bit laughable from a UI point of view, is the redundancy of the “Home Directory”. In this dialog, it is accessible from three different locations as you can see from the screenshot. Also, KDE should take more initiative and add more “Favorite” places along wth the “Documents,” “Temporary Files,” places: My Music, My Downloads and other options as found appropriate.
The Encoding Selection should be under the “Extras”. The “Detailed/Short View” should be under Extras/View as well. The Preview does not work if one has selected “Separate Directories” from the Extras/View menu. This needs fixing. The Favorites MS Outlook-like bar on the left, need to have a little button like Mozilla has, that it will allow it to autocollapse. The “Preview” button should go on the left or not be there at all. A Preview should appear immediately after one has checked the “Automatic Preview” checkbox. The Preview button is redundant there, and in fact, it confuses the user as it does not even align correctly with the “OK” and “Cancel” buttons underneath it.
The context menu inside the dialog is really poorly designed. When someone has selected some files or directories inside the open/save dialog, almost the same context menu as the one described above for the desktop icons should apply. Instead, what we see here is a mix of options that apply to selected files and to actions that should be done when no files are selected (eg. Back, Parent Directory, Sorting etc). There should be two kinds of context menus there, one for selected files/folders and one for when nothing is selected. Currently everything is in a mix, and nothing is obvious. From usability point of view, this needs fixing ASAP.
Konqueror
To be honest, I am not so sure that having Konqueror being able to do Koffee is a good thing. Konqueror, through its KParts, is able to display from CVS, to FTP, to local files, to images, documents, web pages and much more. This sounds nice. In reality though, it makes Konqueror unmanageable.
There are so many options on its menus that, too frequently, have very little to do with your current action (e.g. web browsing, while I find options on the menus that better apply on local browsing).
1. I want to be able to put the location toolbar next to my icons one, but Konqueror will refuse to do so, even if there is sufficient space for it, because a link inside its History has a long URL making the virtual size of the drop down menu too wide (this is a Qt issue and it needs to be fixed in an API level, I think), confusing the toolbar code.
2. I’ll be damned, but I added a local bookmark through my Save/Open dialog (/home/eugenia), and I cannot see it from the main bookmarks menu on Konqueror. This is not what I call consistency. Even if you have to separate web with local bookmarks through a menu seperator, that local bookmark should be visible from all bookmark menus.
3. Bug report time: If you remove the “Font sizes” toolbar from Konqueror and then you add it back in on the main toolbar, the two font size-related icons are not visible anymore and you will need to restart Konqueror to get them back (this does not happen with other icons/toolbars). I saw there is a bug report for it on the KDE bug database, for months now…
4. Toolbar icons should be by default on 16×16. The current default 22×22 is possibly a bit too big. But this is just a personal preference, I guess.
5. Konqueror does not support “border-collapse: collapse;”. Ok, this is not really a problem, it’s just a good way to ask for support for it as OSNews uses it in its CSS code to make its tables look good. 😉
6. I would like to see KDE have a better integration with the system and be able to have a “Copy Image to Clipboard” from a web page and then be able to paste it on Gimp or in a KOffice document.
Qt/GTK+ Collaboration
And this brings me to the GTK+/Qt issue, which in my view, is an important one.
Copy/Paste between the two popular APIs does not always work. I am not talking about the “X11’s middle-mouse-button paste”, but I am talking about real integration between GTK+ and QT in the API level. It is absolutely unthinkable to use a desktop that launches applications with different look, feel and behaviour. Let’s face it; While KDE has a lot of users, Gnome has a lot of developers. Unix is traditionally C-based, and GTK+ is basically C, while QT is C++. A lot of important applications have been written in GTK+. From Evolution, down to AbiWord, Gnumeric, Mr Project, GTKatalog and a whole bunch of other applications that are forming along with the Qt applications the overall X11 experience. People use both GTK+ and QT applications on their desktop. It is a common fact. And KDE and GTK+ developers should use an intermiediate way of making the copy/paste work well between their applications. Also, the KDE project should create a GTK+ theme for both GTK+ 1.2x and 2.x that is able to somewhat mimick the KDE default theme (colors, widgets etc). The Gnome guys on the other side, should also do the same when they are loading a KDE application and make the KDE application look like a GTK+ one. While differences will still be visible (a theme can never replace a real API/coding integration), but at least it will give a good feeling of consistency on one’s desktop. This “trick” will not only benefit KDE, but Gnome as well, and most of all, the user. And it will help bring Unix a step closer to the desktop in the eyes of the Windows or Mac user who are used to a consistent desktop.
The Looks of the Default KDE
I would like now to take the time and discuss the last three issues I have with KDE 3. While there are a lot of UI clean ups to be made AT ALL levels of KDE (from the system panels/apps down to the included applications), it would take me many hours to analyze all. I would leave the rest of the work for the UI designers of KDE. I just hope that the developers listen to the UI designers, because it is known that developers in general, do not always want to listen to the “artists”. This is a common problem on all co-ordinated open source projects. Developers are coding “for fun”, and take orders by no one. Anyways, this is a completely different issue.
1. Important! The menu entries on every KDE’s applications are extremely close to each other. Give it 4-5 more pixels please! The new Gnome 2.0 does it lovely and correctly in this respect.
2. I hope that donators or even TrollTech themselves will be able to buy some high quality fonts for KDE that are able to AA by default and that are able to installed to the system at the same time as KDE is installed. That Helvetica font just doesn’t cut it anymore in the year 2002.
3. Important No2. The current default theme for KDE is butt ugly. I personally use a (modified) .NET theme that comes with Qt. Even the .Net theme I chose to use in lack of something better, has extreme flaws (no grab bars below windows, extremely flat-looking, the blue selection color is ugly – use a lighter/softer one, the grey color on the left of each menu is too dark), however, it is still better than the default KDE one. I trust that the KDE folks have something better to offer in their future versions. By adding the Crystal icons, maybe SVG support, something as nice as the *modified* .NET
theme I suggest or something completely fresh and interesting looking, would be a welcomed surprise.
Please do not forget what we said about the defaults. If the defaults are not nice or what I consider “right,” they can immediately put off a lot of people. Most users do not care about modifying and changing appearance. The defaults have to be right, and the default KDE overall theme is looking old, tired, and uninteresting.
Conclusion
In the beginning of this long article I quickly spoke about integration. Now that you have read all the above, you might think that if the KDE folks add all these suggestions in the next release, Eugenia (that’s me) will be happy and she will (at last) be convinced that KDE is “ready for the desktop.” I sincerely wish that this would be the case. Unfortunately, it is not. What every X11 desktop environment needs is both polishing and integration with the underlying system. Joe User does not want to type arcane Unix commands to install a new driver, or modify this or that. No matter how easy is for you and me that “./configure; make; make install” it sounds like jibberish to Joe User who recently came from the Windows land. Things how to change the refresh rate or the resolution on X11 (please, don’t give me that ALT+CNTRL+(+)(-))? How to change your graphics card driver? How to change the driver for the sound card? These are all problems that KDE should solve through a GUI solution, no matter if the KDE project thinks that it is not their job to do so.
KDE is today the leading X11 desktop environment, there is no denying in that. With great success, great responsibilities are coming as well. In order to be compared to WindowsXP and MacOSX, it needs to do solve some or all of the above problems. A standard way should be found, that is distribution independant (or, alternatively, distros should follow a standard), so KDE would be able to create such tools. These tools are important for the end user. You can’t and shouldn’t just use DrakConf, Yast2, or apt-get, or portage. These are just tools for four distros, tools incompatible between
them! There are more than 10 important distributions today! What about these rest of distros? There has to be found a unified way, a way that works universally to all these distros, a way endorsed via the desktop environment.
For this plan to work, Linus needs to make some hard choices as well. The Linux driver handle should be made more plug-n-play, more flexible, while there should absolutely not be any driver binary incompatibilities between different sub-versions of the same version kernel (nVidia users have a taste of these headaches). Also, all this library dependancy hell needs to (somehow) go away. Microsoft has solved the issue on XP by including different versions of a library in a single file, and Linux should also pionneer a way that will allow users to move more comfortably around their system and easily install new applications without the fear of dependancy/incompatibility (unresolved symbols anyone, after upgrading to GCC 3.1?).
I am eager to read your comments and opinions on how you think your favorite X11 environment can better itself. What are your likes/dislikes and personal preferences? I certainly took my time this afternoon to explain how I think we can better the Unix desktop. Please use the forum provided and comment on your preferences.
I like the ideas proposed even if the one that actually interested me most were the one about the integration or lack of it. I actually wrote something on KDE to ease of the use of portage and then after thinking about it the problem was really the one mentionned by Eugenia, why wouldn’t I be able to use single app in a single way to install most kind of software.
On other OS, the native way is use as it fit the best the desktop but this is when the os comes with a desktop solution. Linux is special because the application space without UI is great and is well populated compared to the UI application space.
Therefore Linux UI need to be even better than other OS UI to properly presents all the different app available in a consistent way and all that in a fraction of time it actually took to make those non-UI app.
This is what I would call a Monk’s job 😉
Under X (X-Darwin in fact) My desktop of Choice is http://www.GnuStep.org with windowmaker.
Gnustep will get a boost in its developemnt tools. Apple is really working hard on gcc and has started not so long ago to work on GDB. Apple uses Objective which is what gnusteps need ….
Not to re-open old and unanswerable debates, but I feel that X has to die for Linux to make it to the Desktop.
Or at least we need to have some kind of optional replacement that doesn’t carry all the baggage that X does.
A lot of what you point out in your article makes sense Eugenia, but even if the folks at KDE did make your suggested changes (not likely… There’s a huge difference between what OSS developers want vs. what their users want), I still don’t think Linux would make the inroads that it wants to, into Desktop land. KDE might be ready too, but it’s underlying foundation is not.
X lacks the easy configurability and universal performance that a desktop OS needs.
Yes, you can login remotely to X. I know, I know… That’s what all the die-hards chant whenever this debate comes up. But how often have any of us done this (System Admins don’t count)?
Ok.. Now compare how many times you’ve NOT used X’s remote features, as compared to how many times you’ve wished you could change your resolution easily without rebooting?
And this is the crux of the arguement: Yes, X has some very unique and arguably nice features. FOR SYSTEM ADMINS!
That only serves to reinforce the notion that Linux is only suitable for Server and Legacy applications.
Yes, you can setup some nice dumb terminals and log in remotely to an X-windows system remotely to do your work, but again… Is this what the average desktop user wants??
Not likely. In fact let’s see who would want to configure Linux for this purpose… I know! Again, System Admins!
Yes, some use it as their sole desktop OS, but they are the exception. I want to see Linux have the universal appeal and performance that makes the a modern OS what it is. Linux is trying to move there, but until the Linuz zealots get off their soapboxes and admit that X might not be the end-all to everyone, Linux is staying right where it is. Locked into the server/hobbyist niche that it’s carved out for itself.
And the funny thing is that although there’s tons of people shouting for an X replacement, very little headway’s being made on creating one.
So while KDE and Gnome both have grown up nicely (I agree KDE’s leading the way though!), they’re both also tied to this old behemoth of a display system which pales in comparison to their sleek, high performance GUI’s.
While I appreciate the work going into both GUI’s (BTW, XFCE’s still my preference, even with KDE’s recent improvements, largely due to the performance hit that X gives KDE), until X is gone, or there’s a compatible alternative with better specs and performance, I don’t foresee either taking over the desktops of the average PC user.
Sad too… KDE does have some nice features. Until you want to interact with something that’s not a KDE app that is. Then it’s a mixed blessing (again, largely due to the lack of a universal, high-performance backend, keybindings, and so on; All of which a good, well-written X replacement would bring to the table.).
Ok X Lovers: Flame away…
I Definitely appreciated this look into kde’s I design. I hope they listen.
Thanks for that, Eugenia. I hope you sent a hint on that to the corresponding KDE mailing lists.
However, allow me a one remark: Why do you reduce the width of your panel? it would make much more sense to reduce the height and increase the width over the full screen width. Yes, kinda like Windows…Microsoft did that for a reason. With the screen corners and edges being the points that are the easiest to reach, take advantage of that and put important objects in the corners and at the edges. In this case, Kicker in any screen corner and the task bar (sorry, I don’t know what the KDE name is) and a border, and don’t stack items in it! The two-row layout takes away everything you gain from placing it at the edge.
But I do not reduce the width of my panel. In fact, I make more use of the desktop space!
Everytime a new application opens and creates a new Taskabar entry, Kicker needs to *resize automatically* to the required width (+150 pix)! But today, Kicker does not ‘listen’ to Taskbar’s size requirements. This needs fixing. If they fix that, it is better to have a “dock” like OSX’s, and then, as you add more applications, the width will expand to eventually across the screen, as it is in the default setup of Kicker!
But why do you make it use so much height? Why two rows? Why don’t you keep the K menu in a fixed location?
I like it better this way. The K menu otherwise is extremely small (16pix) and everything looks too flat, uninteresting, and it gives the impression that Kicker is of no importance. Which of course is not the case.
Even the XP menu is a lot “fater” these days in terms of height than it used to be. Same goes for OSX’s dock default height.
You don’t need any height when your at the bottom. But you are right, the K menu neeeds some hightlighting. I think just giving it some colour would help here.
Great article, with tons of interesting points. I hope someone listens. The point about default configs is quite important.
And yes, a bit more speed will be nice as well, though I am not sure this is entirely a KDE issue.
> And the funny thing is that although there’s
> tons of people shouting for an X replacement,
> very little headway’s being made on creating one.
Attempts have been made, but they all remain half-baked. Wishing for a replacement is easy, but writing one, and actually getting it to the point where it is useable, (and able to support existing X apps), is an incredible amount of work. I was hoping that KDE would actually create that replacement eventually, but judging from past interviews, that possiblity would seem to be remote.
Anyways, great article, and kudos to the KDE folks for doing a great job. I hope KDE gets better from here.
Would it be too much to ask for a Linux distro made specifically for KDE so that it may be optimized for KDE and really integrate KDE into the underlying system?
Turn on autohide if KDE is capable of doing it. Then you’ll be taking up NO space for the taskbar, but when you need it, it pops back up with all the room you need and then goes away.
1. People love to go on and on to say why X should die. I used to be one of those people, until I wised up. All the reasons that Mr. C gives has to do with the implementation, and are not inherent to X. People inanely chant “X11 sucks! X11 sucks!” just as much as apologists chant “Remote apps!” There is no reason that a system like Mac OS X couldn’t be implemented on top of X11. You could implement Display PostScript (like with GNUstep) or Display PDD (OS X’s quartz) on top of X11.
With straight up vanilla X11R6.3, no. It lacks transparency. But there are hacks that deviate that bring transparency to X. We all could be running this X server without it ruining our old apps. Organizing a big push to get all the major distrobution organizations to start using a non-standard X server would be a lot easier (and more practical for all involved parties) than would to have it Berlin (or something akin to it) replace X11.
Your gripe is with Xfree86 and the existing toolkits. The way you change resolution and bit depth is specific to the implementation, in this case XFree86. You do it different on Solaris under OpenWindows. Before you open your mouth to chant the “X11 Sucks!” mantra next time, remember this. Change it to “XFree sucks!” if it’s that important. These problems could be fixed within XFree86 itself.
The culture of X11 is somewhat icky. There is no built in widget set- everyone and their grandma has written a new widget set that doesn’t work with the other ones, look slightly different, but does pretty much the same damn thing. The way X is, and the fact that Xaw, Xt, and leter Motif suck hard encouraged the emergence of all the new GUI toolkits.
I’m not so much an X supporter as an enemy of ignorance. That is, I’d much rather use something like NeXTSTEP or PicoGUI (http://pgui.sf.net), that has a widget set built into the display server.
Toolkits. The source of many other problems. KDE and GNOME lack integration with themselves. The integration you find in Mac OS. From my observations, most KDE and GNOME people come from a Unix or Windows background, two platforms where there’s not all that much consistency and integration. Sure, a shell brings a shred of consistency, but not as much as the shell in VMS. (no, I’m not kidding, do your research before laughing) In Windows, whatever integration and consistency that is there is easily missed, or is convulted to the point that these KDE and GNOME developers must be getting the message that any sort of attempt at integration and consistency just gets in your way- as it often does on Windows.
The Linux OS has a little bit to do with a lack of consistency and integration. However, NeXTSTEP and OpenStep has shown us that a usable *user-oriented* environment can be done on top of Unix. There has to be wrappers around the old interface and some padding, but it can be done. Doing that kind of work is not fun though, oftentimes, and frankly, I don’t blame Open Source and Free Software developers for not wanting to do dirty work they see as unnecesary. I am a OSS developer myself, but I must admit sometimes I would much rather just skip to the fun stuff.
It’s too bad OS News doesn’t have threading, or some way to reply to a specific post. I know that this large editorial will probably go mostly just be drowned in the chorus of yelling, but if anyone wants to write a reply or rebuttal, an email would be welcome.
I totally agree with TLy. There are many strengths in having a generalized distro- but we already have 100 of those. What would be incredible are distrobutions that take a desktop, KDE, GNOME, GNUstep, XFCE, and take the desktop to the next level, basically an OS entirely around KDE. Get rid of all of the Unix cruft that you can, make administration as easy as (bleh) or easier than Winblows, and have it well integrated with the Linux or BSD install below it. Hell, it would be cool if someone created this layer- what goes between the traditional Unix/Linux side and the desktop- and had it so that pretty much any desktop or app could call into it. <sigh> I should just asking for OS X, and switch back to it.
TLy : “Would it be too much to ask for a Linux distro made specifically for KDE so that it may be optimized for KDE and really integrate KDE into the underlying system?”
i totally have to agree with that, maybe by this way we could gain some speed and stability:)
There is such a thing. It was called Corel Linux, will be Xandros Linux soon and is also providing the base for Lindows.
To the guy saying how bad X is: I agreed with you but not anymore because I learned that performance isn’t the only thing that counts and X can be made quite responsive, also looking good. So what’s missing is mainly configurability and I believe that it’s definetly possible to add this to X (might take some time though but not as much as writing a new display manager from scratch). Of course there is nothing wrong with writing something better, but currently I believe that X will do the trick for now.
To Eugenia:
About Gtk/Qt, you are a bit wrong with stating that it would be common to run Gtk and Qt apps at the same time. I never do that. Both environments are really mature enough to offer whatever you need. If you don’t want to use Gtk Evolution, use Qt KMail. Sure, you might like Evolution better but it’s still your decision, you don’t have to.
We have to understand that Unix isn’t just _one_ desktop environment (and now please don’t respond with an upset “that’s why Unix will never make it to the desktop”) but many. Just like we have Windows, Mac OS and Amiga OS, we have KDE, Gnome, XFCE, etc. All of them can live on their own, some have more applications, some have less. What they all share though, that is X and the ability to run every application if you _need_ it. That’s a great advantage that some proprietory DE’s don’t have. Those applications will never really integrate (this might be sad but it’s true), but they will work together. Things like clipboard, etc will work (you said it doesn’t work for you, is this still true with Qt3 adn Gtk? Qt2 had a buggy clipboard implementation).
BTW, most “Joe Users” I know have much less problems with the inconsistent looks of Gtk and Qt than I have. For example my father runs all kinds of different applications and he really doesn’t care. For example a while ago he had a Go client on Java with this ugly old Motif toolkit and he told me it would look pretty.
So it’s really not always about the toolkit, especially not when it comes to “special task” applications (anyone know Blender?).
About the UI of KDE, I mostly agree with you (I especially agree that Konqueror doing everything doesn’t make him an actually good browser and with the default style beeing bad. Mosfets Liquid is very good, but also very fancy…). Your suggestions are quite reasonable. Although that’s really not what makes me shiver when sitting in front of a KDE desk.
My problems with KDE are those:
– Cluttered menubars. As you pointed out, they don’t have much space between menu items and I think they have a reason for this.
I usually don’t get the menu structure of a KDE application, just like I usually don’t get the menu structure of a windows application (like Office), so when I need a certain option, I’m usually searching through the menus instead of immidiatly knowing where to look. :/
– A cluttered control center. Same problem as with menubars, when I want to change a certain feature, I usually have to search a while. They just have way too much preferences and I agree with Havoc Pennington’s idea that too much preferences are the root of all OSS GUI evil.
– They also copied other UI mistakes from Windows, like MDI. Granted, this is mostly optional so it’s not really a problem. But I find many UI parts of KDE to be build with functionality in mind, not usability. So I can do a lot of things, but don’t do them efficiently. Example for this is the Konqueror “split view” functionality. It’s a great idea, but I could never figure out how to efficiently use this.
– The default kicker settings don’t come even close to what I would consider to be an intuitive and efficient desktop interface.
And some other details. Even the tiny detail that “Ok” buttons are shown on the left makes no sense to me anymore. Some people bash Gnome for breaking out of the “norm”, I comment them for having the courage to stand up for what they believe in. Those little details really make me enjoy computing, even when I’m used to something else (less intuitive).
I also agree that system tools should be embedded into the desktop. At least the frontends should be provided, so that the backends can be manipulated for certain systems. It should be the responsibility of the system creator (distributor) though to fit everything together so you get exactly the tools you need to manage this certain system and to make sure that they actually work. BTW, the XST Setup Tools (from Ximian) are a very interesting project that could deliver something like that. It would just need more people actually using and supporting them.
> Mosfets Liquid is very good
I truly hate Liquid. It is way “too much” at all levels.
Anyone know where I can get debs (prefebaly, just by adding something to my sources file) for KDE 3 *or* GNOME 2 for Debian GNU/Linux 3.0 (woody) PowerPC? I’m dying to try them out, but don’t want to take all the time to compile and futz with them, or to break my lovely packaged system.
This was a good article, but I’d bet my bottom dollar that most of the advice will be ignored. I’m beginning to think that Linux will *never* be ready for the desktop – there’s far too much historical baggage. In no particular order there’s the multiplicity of desktops, the lack of standardisation, the software package management situation. Above all else, however, is the lack of design. This article makes that lack very clear.
We all know that most of the developers who write free software do it in their spare time and work only on things which please them. When you get a multiplicity of developers adding “cool” features you ultimately wind up with bloat, with many different ways of doing the same thing and with no standards. That’s why Linux on the desktop is doomed.
The thing that keeps me on Windows is that windows doesn’t get in my way. I find that KDE and GNOME are consistent in getting in my way when I’m trying to do my work. If I want to play, to fiddle, then I can fiddle to my hearts content – especially on KDE with the many different combinations of changing the look (but not the feel) of the user interface.
The only thing GNOME has going for it is Ximian, where people are paid to pay attention to the types of issues highighted in the article.
At the end of the day, it’s a question of usability. If I can’t use my computer as a tool because I’m constantly having to fight the interface, then I’ll have to use something that allows me to use the computer as a tool and at the moment that something is Windows. Unfourtanately, it’s the old 80/20 rule, only in this case Linux desktops are stuck on 80% and Winblows is more like 95/5. That extra 15% adds up to usability and until KDE and GNOME get there, linux won’t be ready for the desktop.
IMHO of course…..
…by that I don’t mean that it’s not a good thing – everyone likes it when things look similar. But inconsistency in UI has not prevented some technologies from becoming dominant.
A few examples:
the WWW (few sites have the exact same navigation, even though trends can be definitely seen)
Video and PC Games (each game has its own menu structure, look, and different use of controller interfaces)
Even Microsoft isn’t consistent in its interfaces: look at the difference in menus, toolbars, dialog boxes, etc. between Win95, Win98, Win2k, WinXP, Office 97, Office 2k, Office XP, etc. That’s not counting other applications…
So, yeah, consistency is good, it’s a plus, but it’s not the holy grail that will help Linux on the desktop.
Hmm. I remember a certain review of Gnome 2… where Eugenia said that Gnome 2’s defaults weren’t good, and didn’t even bother to take the effort to realize that the menu panel is easily removable. Then, she gives a two-page summary of how KDE should be changed to make it better. Why not give Gnome the same treatment?
Nothing personal. It was just that the Gnome article came 2 days after the release of Gnome 2. It was a review.
This KDE article was under writting for the last week. It is not a review, it is a feature article.
If I am to write a similar article for Gnome, it would take me much more time. Time that I do not have. Gnome needs even more *basic UI changes* than KDE needs, everywhere in its system. All the developers have to do, is listen to their three UI designers.
After they fix the basics, I promise to do a feature article for Gnome too, that would discuss the further polishing of the UI.
http://www.hadess.net/
should help you getting Gnome2 on Woody/PPC. I don’t know where to get .debs of KDE3 though.
WWW sites and games are individual productions.
The method of accessing multiple sites is the same – ie you use one browser where the navigation buttons are in the same place etc. The navigation UI doesn’t change every time you visit a new site.
IMHO of course…
“The only thing GNOME has going for it is Ximian, where people are paid to pay attention to the types of issues highighted in the article.”
Yes, and people from sun and redhat and…
It really payed off IMO, so far I agree with every Gnome 2 design decision and it’s just a joy to use. Just like you said, KDE and Gnome 1.x where great in getting into my way. :/ They made me fiddle around with the system all the day. That was cool and all but I have to live from something. =) So I used Win2k for several months, looking for a free desktop alternative. While Gnome 2.x is “less”, I would definetly say that’s it’s biggest strength. It’s great in not getting into my way.
I’m really looking forward to the first matured versions of Gnome 2 (like Ximian Gnome 2.2
), because many people will be surprised… I’m sure of that. And this will set the bar for usability just the tiny bit higher that is needed.
“the WWW (few sites have the exact same navigation, even though trends can be definitely seen)”
Very good point! I still remember the time when I was pissed of Windows because everything looked the same. I wanted original applications and consitency was something that rather annoyed than pleased me. I wanted my system to launch applications, I didn’t want the system to be everything I use all day long. So I agree with you that consistency is really overrated. Maybe we are looking in a completely wrong direction. The important thing is, that it _works_.
“If I am to write a similar article for Gnome, it would take me much more time. Time that I do not have.”
Stop the trolling.
There isn’t really much GUI about Gnome, so even if all if it would suck, it couldn’t take so long to write about…
stew:
Alas, no dice with that. According to the Hadess | Idoru page one must “Be running Debian UNSTABLE. Not Woody, not potato, only unstable” to use their GNOME2 debs.
Aaron
All the developers have to do, is listen to their three UI designers.
Do you mean
a)they had 3 UI designers since the beginning, but they didn’t listen to them until the complains about Gnome2’s UI appeared?
b)they got 3 UI designers after those same complains?
As I know you talked with some of the Gnome2 about those problems, does it mean they listened to your advice and noticed they needed to get good UI designers?
Anyway, I hope you keep making these kind of articles, it sure helps development of open source desktops aimed at users.
UI…
User Interface
Personal Preference.
again, linux isn’t windows, windows isn’t linux.
and one person writing 90% of the “articles” opinion, regardless of the pat’s on the back by the regulars, doesn’t
mean, well, anything.
linux was not, and to my knowledge right now, developed to replace windows on the desktop.
these are all very recent events. so is the concept of
“UI” design standards for a graphical interface for
a computing workstation. i mean really, think about it for more than two minutes or whatever time it takes to make yourself feel special or talented. stuck in the age of the wheel vs. stuck in the age of the taskbar.
yes, feedback is *great* and plentiful options for one’s
workstation / enviroment is *great*
everone has and has the right to, an opinion.
but a little perspective, and less personal attachment.
i mean for the love of god, we’ve watched people prattle on why their platform of windows/macintosh were superior for how long now? shall we continue now with the stupefying
differences between linux distributions? while maintaining
the decade plus old debate as well?
i know i’m wasting text, but it’s just plain
sad.
but hey. we’ll all keep busy. that’s the important thing.
apparetnly.
Eugenia – I read and commented on your gnome2 article and the basic gist I remember was moaning about the themes and fonts – aside from the fact that the install was obviously borked – most of the comments refered to the default layout, so the comment above about fairness in reviews stands.
Personally I think the KDE interface is horrendous.
I’ve always preferred KDE over GNOME but the GNOME panel is far ahead of kicker.
What I’d like to see in kicker is:
* Ability to have vertical separators between icons
* Kmenu button should automatically have this border on the right/left
* Ability to make drawers a la GNOME
* Did you ever notice that when you have kicker at less than 100% width, you can’t activate the desktop menus even though the desktop wallpaper is displayed in the unused space. This is extremely unituitive.
* The ability to have a tasklist which doesn’t have text in it, i.e. use the app’s icon at whatever size kicker currently is at.
* We need separators in KMenus, too.
BTW Eugenia, I absolutely love having the run menu on the root Desktop menu. This is one of the reasons I choose KDE over its competitors (though WM has it, too). It is one of the most frequently used commands from the desktop menu and it makes no sense to shunt it off into a sub-menu. What would be nice is if everyone could customize the desktop menu to one’s liking.
> I absolutely love having the run menu on the root Desktop menu
This wouldn’t be necessary if the GUI was intuitive enough and if many system services were able to be easily carried out through the GUI. The “run command” menu under an add-ons menu makes more sense, and it would fit better there, if a whole bunch of that kind of utilities are there too.
Check BeOS’s Tracker add-ons to see the kind of utils that people would be able to add on their add-ons dir!
http://www.bebits.com/browse/29
This is probably one of the best KDE reviews I have ever seen. I agree with everything in it.
kirby, there isn’t much of anything that makes my blood pressure rise when it comes to talking about OS’s because it is fun to talk about them. But, when someone spends quite a bit of time to write a feature article dealing with something that, I’m sure, the vast majority of us here are very interested in and then to have smug replies such as yours (with bad spelling to boot), that does get my pressure up a little bit.
To me, when it became obvious that Be was going nowhere, the idea of Linux for the desktop became the most exciting thing in computing. It is obvious that many other people feel that way – that’s why we talk about it.
This situation is so different than the history of the major desktop OS’s – with Apple and Microsoft, for example, they poured who knows what kind of resources into propping up OS’s with weak underpinnings, until they finally had to totally move to something with a strong foundation. Here, we have a different situation – we have the underpinnings. We have the kernel, we have X. The foundation has already been laid and it is important to criticize, review, suggest and hash out how this can be made better. If you are bored by all of that, that’s your right. But, if people had the attitude you do, we would, indeed, still be in the age of the wheel.
That title didn’t meant for KDE. I use KDE most of the time on Linux, however I just want to mention again about EDE (Equinox Desktop Environment).
In my opinion this DE got the concept of user friendliness, responsive and fast. However the look is quite ugly. I think if KDE could address the issue of crowded menu items, slow responsiveness and long launching time (including application) it will become the best DE for Linux.
If they don’t, other DE will catchup and I think EDE could do it soon. Want to know about latest EDE? Just go to http://ede.sourceforge.net
The article states that normal users should not be allowed to use the quick browser for the / directory. Why? This would be just plain annoying. If they are allowed to acces a directory, then they are allowed to access a directory. If they aren’t, they still aren’t. Try browsing to some directory you are locked out of, and you get Failed to read directory, which works.
I also disagree with the replacement for the menu tear-off option. The dotted option may be archaic looking, but what the article suggests as a replacement is not clear as a tear-off option. The only reason why I knew what it was is due to the direct comparison to the tear-off dots. Perhaps a jagged-edge to look like torn paper? Also, having it show up only at an OnMouseOver would mean that many people wouldn’t even know it is there.
Finally, I disagree with this being posted as an actual usability assessment, as opposed to the personal opinions of someone who just happens to be a UI specialist. A proper usability assessment allows several “Joe User”s (to use the author’s rather perjorative term) to attempt to perform tasks, and records reactions from them. With 50 — 100 people attempting to perform tasks with a system, mostly (if not all) people who are unfamiliar with the product, one can make halfway-decent inferences.
Equinox Desktop Environment? I didn’t know about this one. Anyway, that’s not ugly, it looks similar to Windows 9x (the UI elements) and also to KDE (the icons on the desktop). I didn’t know about this DE so thanks for mentioning it. I looked at the screen shots, it looks amazning! I am so excited, I want to try something new, I am gonna give this DE a try and hopefully I will keep it. Whenever I use KDE, I get sick of it, as I said, I agree with all the things pointed out in the article, and KDE always makes me go back to Windows.
http://www.equinox.szm.com/screenshots.html
Some interesting UI ideas, but I’m more interested in the OS integration issues mentioned:
Linus has *no* work to do about nVidia drivers – I have a nVidia card on one machine, and it’s a PITA. Definitely not a Linus/Linux problem, though, just a p*ss-poor driver from a 3rd party. That’s comparable to saying, “Windows sucks, it crashes whenever I plug in <some specific hw item>”… 3rd party driver problems are not an issue for the UI or OS teams. nVidia just need to catch up.
The other biggie mentioned is apt-get / rpm / etc… again, this is not KDE’s problem to solve, for the next year or longer, this is *not* going to be solved. Sure, from the Linux<==>JoeUser point of view, it’d be nice, but getting political consensus from all the different distros (and their users – many of whom are well-paying customers, who are tied in to .deb, rpm, etc) is not something I would wish upon my worst enemy.
These “integration” issues sound more like “isn’t it nice when everything’s controlled from a single source (eg MS / Apple) … the Linux desktop should be like that”. Linux would not *exist* if it were like that … as the author acknowledges, the development is mainly done by volunteers, who (like it or not) would revolt en-masse were a single entity (person/company/whatever) to start dictating how the kernel, UI, packaging, etc should function.
Linux *isn’t* going to be like that.
The environment required for what the author asks, is: a single non-benevolent dictator who spends 90% of their time listening to usability experts, 9% on interaction with developers, and (let’s be charitable) 1% left for actually keeping up with the technology.
The developers must then, in this environment, have some reason to reject superior technology in deference to the dictator and focus groups.
If an OS could be developed in this way (and maybe it could, who am I to say?), I would certainly not be interested in using for either desktops or servers.
> The article states that normal users should not be allowed to use the quick browser for the / directory.
Not via the desktop menu, no. Remember, this is a multiuser environment, and users should stay at Home (no matter if no one does at their home installations of Linux). The root user should be able to do user /.
> but what the article suggests as a replacement is not clear as a tear-off option
Tear off menus is not a great idea altogether for all the users (although useful at times), and therefore it should not be encouraged as much. My replacement was a 5-minute idea. Point is, it needs replacement with mine, yours, someone’s elses… Thing is, it looks truly ugly as it is now.
> I disagree with this being posted as an actual usability assessment, as opposed to the personal opinions of someone who just happens to be a UI specialist.
I think you did not read the article. Please don’t get me overboard, and I am in a pretty good mood today.
This IS my personal opinion! Read the first 3-4 paragraphs.
> A proper usability assessment allows several “Joe User”s
Yes, but I am only one. And if I count my husband, we are two. That’s all I can offer for now. When we get children, I will make sure that I will publish a “usability assessment”. Duh.
(wo)Man, It could’ve been me who wrote this article.. I’m feeling exactly the same way about it
I really hope the KDE developers and artists will use these comments to improve KDE!
I agree with everything you said about Kicker. To me, this is the most under-improved part of KDE 3 – it doesn’t seem like anything has been improved on it recently. I also turn off the zooming.
What I would like to see most though is making editing the K menu (and the Kicker quick launch icons as you mention) a whole lot easier. Drag & drop/editing on the menu itself (like the Windows start menu) would be nice, for starters.
> Linus has *no* work to do about nVidia drivers – I have a nVidia card on one machine, and it’s a PITA.
You misanderstood what I meant with the nVidia example:
1. Linux drivers are losing compatibility with each sub-version of the same version kernel, making binary/object drivers like nVidias to require either re-compilation for your kernel, or to download a version made specifically for the sub-version of your kernel. That is a big problem for the third parties and it should be solved in the Linux kernel. The Linux kernel SHOULD NOT lose binary compatibility with itself every 2 months. It is a big problem for the OEMs and the developers!
Read this Joen On Software on the topic as well:
http://www.joelonsoftware.com/articles/StrategyLetterV.html
Secondly, I will describe my personal problem with the issue:
I have Red Hat Linux here, and I have three kernels booting on and off. I am a developer too, so I get to test stuff and play with new stuff as well. So, we bought that nVidia GeForce2MX400 card to replace the onboard crappy SavagePRO+. Problem is:
1. I have to recompile the nvidia drivers THREE times, for my THREE kernels.
2. Problem is, I have upgraded to these kernels 1.5 months ago, and since then, surprise, I upgraded to GCC 3.1. Now, the nVidia driver REFUSES to compile itself, because it says that the kernel was compiled with a different compiler, so I will need to either downgrade my compiler to the same one I did the kernels (GCC 2.95.x), or recompile all my kernels with GCC 3.1, in order to retry compile the nvidia driver.
These my friend, are not nVidia’s problems. Are Linux’s, library hell, gcc etc etc etc. The whole damned system’s on how it was created and maintained.
I know, I know, it’s not what you asked for, but here’s where you can find KDE 3 debs for debian unstable/sid:
http://calc.cx/kde.txt
I haven’t tried it btw.
biggest problems in my opinion are:
1. stupid defaults
2. poor drivers
Although KDE is now very feature rich, graphics is still very flat.
OSX, Windows are IRIX could not be seriously compared to KDE or Gnome, until free alternatives will concentrate on basic graphics instead of new featues.
#kde3 beta sources
#deb http://kde3.geniussystems.net/debian ./
deb http://kde.debian.co.nz/debian ./
deb http://kde.ping.uio.no/i386 ./
#deb http://people.debian.org/~schepler kdegames/
deb http://people.debian.org/~bab/kde3 ./
#deb http://people.debian.org/~njordan kde3.0/
I definately like your suggestion for the rip off menus. It looks a lot better. I really liked the Mac’s way of just holding down the mouse while mousing out of the menu to rip them off.
I also like the first screen shot where you have the big K and pencil and have the other icons small in Kicker. That should definately be the default look. Is that possible to arrange it that way in KDE2?
I like having the file type show up by right clicking on the file instead of having it under properties but I do think they should combine some things. They should combine the screensaver and configure desktop.
In talking about wording, I think the desktop menu should say “new folder” instead of “new directory”.
I also wish they would have Kpm as an option when right clicking on Kicker. I use this constantly in windows and in KDE.
The two things I hate most are copy and paste and moving files!!!
If I copy something and then highlight something else, it should not have overwritten my copy. When I highlight and drag something, it should not copy it. It should move the highlighted text.
When I drag a file to another directory it should just move it, it shouldn’t ask me if I want to copy it or move it. 90% of the time I just want to move it. I also wish they had spring loaded folders to make moving things even easier.
Also, find increase font and decrease font should not all have magnifying glasses for icons.
They should also redesign the control panel. They should combine the features of Mandrake’s Control Center, KDE control center and linuxconf. In the KDE control center they should remove the Web Browsing portion and File browsing keeping them in Konquerer’s preferences only.
Looking at new KDE 3.0 and 3.1 screenshots, KDE definately looks a lot nicer.
“These my friend, are not nVidia’s problems. Are Linux’s, library hell, gcc etc etc etc. The whole damned system’s on how it was created and maintained.”
Huh? I wanted to agree with you until I read this sentence. Now what have library hell and gcc to do with this issue? What do you expect when upgrading to a new gcc major release.
The only real problem I see here is the lack of a frozen driver API for (at least) stable Linux releases. Unfortunatly I don’t think this will change anytime soon because Linus and Co constantly state that they don’t plan to keep the API stable. This makes me wonder all the time because I don’t get what’s such a big deal in doing those API changes only for development drivers.
Almost everyone gets this right (even AtheOS), why not Linux.
It would solve one major problem for Linux on the desktop. The only problem that really belongs to the kernel.
“If I copy something and then highlight something else, it should not have overwritten my copy.”
This shouldn’t happen anymore with Qt3 and never happened in Gtk. The way it should be is, when you copy something (with CTRL+C for example) it will be copied to one clipboard, if select something, it will copy to another clipboard. Middlemousebutton will always paste from your second clipboard and CTRL+V, menu, etc will always paste from your first clipboard. Try it.
Qt3 and Gtk both have excellent clipboard support.
“When I highlight and drag something, it should not copy it. It should move the highlighted text.”
Good point, that’s also how it works in Gtk2 (it will be moved). Maybe this would be a good suggestion for KDE’s bug tracking system.
My favorite KDE snafu is under the “System Notifications” section of the Sound kontrol center section. If you look at the available notification settings, you will see that there are 8 different sounds that you can apply to the virtual desktops, while you can have 16 virtual desktops… someone needs to run kalc and do some math 😉
Good article.. I agree with everything except the preview being on the left in the file open / save dialog box.
Have a great day!
To me the best solution would be for kde to have a package installer that does the ./configure; make; checkinstall; (replace checkinstall with something for your distro’s package system) for the user. The user would not have to see the details. The distro’s pakage manager would have to help with dependancies, but it would solve a bunch of problems. A single package could work across many distros since most of the director/library version/compiler differences problems would be avoided. Options could be easily set from the gui and passed to configure.
The problems I can see with this are: speed, and having to write plugins to make it work with all the native package systems. Speed is not a big problem in my mind, since if binary releases should work fine within this system if they are needed. For distro support I think if something the size of kde puts its weight behind something like this I believe the distros will do their part.
I could see this working well for nvidia’s driver at least. One package for all ditros, download, click on it, hit next a few times and you have installed a driver that works with your kernel.
There is a project called DirectFB (www.directfb.org) that basically replaces X. If a lot more attention was placed into getting DirectFB out there faster as an X replacement, it will be one step in the right direction.
The lack of hardware support is the real hassle at the moment. If we tackle DirectFB with the hardware manufactures (ATi, nVidia, Matrox, etc) it will improve a lot faster than what it is now.
On the software side, the Gnome guys are porting the Gnome/GTK framework to DirectFB. No word on KDE though. Maybe it would be a good idea to start on a unified desktop for running under DirectFB. Are the KDE and Gnome developers reading this!!!
I hope this spurs some conversation. Because it could be Linux’s saving grace.
Nonono, something like this doesn’t work. The trouble of compiling something isn’t typing “./configure; make; make install”, it’s figuring out why it doesn’t work.
For example to make the NVidia example work with your kernel, you need the kernel headers installed and you need to compile the module against your installed kernel (so binary packages of kernels aren’t sufficient anymore). Gentoo has this solved quite nicely, because you compile your kernel anyway, they can also compile the NVidia kernel module “on the fly”, so you just have to type “emerge nvidia-kernel”. But not everyone wants to (and shouldn’t have to) compile his/her own kernel.
But maybe it could work if binary releases of drivers would be delivered together with those headers and what else is neccesasry to compile a compatible kernel module. Than a system like portage could compile the kernel module.
Hm. That’s why I think that a portage-like system could solve a lot of Linux problems because Linux is much more friendly to source compilations than binary installations IMO.
That’s one of the most annoying things I find in KDE, the abuse of icons like no other desktop UI does. So many icons get in my way, in KDE icons are like gremlins, it’s an invasion.
Eugenia, I like more your Desktop Context Menu than the default KDE one, I’d like it even more if I was given the option (like in GNOME) of disabling any icons in context menus.
In sort, more or less, I’d like to have the KDE3 functionality within GNOME2 desing.
Design is the worst part of KDE, it needs a completely new, clean and polished interface, preserving all the fucntionality achieved with KDE3. I don’t expect that change from the KDE team, that would be asking too much having in mind they do it just for fun, or so they say. But I sincerely hope the change is brought from companies who put their money at risk like the coming Xandros Linux. As a matter of fact, I wished KDE would stop being mainly a hobby project and companies like Xandros and United Linux take on.
Windows95 looks better than KDE3, and that sucks. It isn’t about eye candy, it’s about interface desing. The KDE team has made a great achievement taking the GUI to do what it does given the lack of economical resources. Now it needs a modern interface, a clean one. Themes are something else.
I think this newer version of KDE is looking better than the last 3.x version. The bottom bar seems a lot less cluttered. It looks a bit more interesting. Personally, I find KDE too slow on my computer. I much prefer a window manager only like WindowMaker or Enlightenment.
Hmmm. DirectFB looks interesting. What about Xi Graphic’s accelerated X servers? Are they also a good extension to X, I’ve never tried it before.
As to replacing X altogether, well thats a novel idea but the purpose of X was to make and extensible GUI for all system architectures running Linux and Unix. Seeing as it was made in the 80’s, I think its still doing a great job. If you scratch it entirely now, you’d either need backwords compatibility or new apps altogether which is a coding nightmae.
I love the article but it has one flaw. The suggestion to remove X because no one needs its features. To be a successful replacement any new system should have the exact same capabilities as the old one before adding new ones. Right now only administrators use X log on because only they usually have multiple machines to work with running Linux. Even when home users have several computers it’s usually one Linux one windows etc. Lets say in the future Linux starts getting a 30% market share on the desktop wouldn’t it be nice to use your old 800 MHz computer X logged int to a brand new 3000? MHz computer and have two amazingly fast computers. That is one of the strongest and most beautiful points of Linux, Nothing gets wasted. Don’t advocate throwing away some of the features that make Linux so great just for short term gain.
Last I checked DirectFB was mainly for embedded devices. I know that Gtk2 runs on DirectFB, but Gnome? Where did you hear that? I don’t believe that this is true until I have proof.
What could be interesting though is XDirectFB, that’s basically X but runs on DirectFB. The advantages are that it starts up MUCH faster, doesn’t have this ugly “X” background, features alpha blending (you can make inactive windows fade into transparency for example) and features hardware accelleration. It’s very premature yet though and crashes for me everytime I load a Gtk2 application. :/ TWM, Mozilla, etc work fine though (didn’t try KDE3). I also couldn’t get my framebuffer to run at 1600×1200 yet…
> I love the article but it has one flaw. The suggestion to remove X because no one needs its features.
Ermm… the article does not say anything about removing X. What are you talking about?
> Yes, you can login remotely to X. I know, I know…
> That’s what all the die-hards chant whenever this debate
> comes up. But how often have any of us done this
I use this very frequently. (To fix other peoples’ computers, to test code on specialized machines, to access the computer lab from home)
However, I don’t think that remote login and network transparency should be stumbling blocks for an X replacement. Looking at things like VNC (and TightVNC, an enhanced version, available at http://www.tightvnc.org), these use an RFB protocol (Remote Frame Buffer), which seems much more flexible. Clients and servers are available for a multitude of platforms.
I use it on both Linux and Windows. I can even remotely view hardware accelerated DirectX applications, because it’s occurring on the frame buffer’s level. (The only thing I’ve found to not work are overlays)
This technology eliminates the need for X’s networking features. At this point, I see backwards compatibility as the primary problem facing an X replacement.
> But there are hacks that deviate that bring transparency to X.
What are the hacks and where are they available?
Now I’m getting a little bit excited over here. It seems that they are serious about it. When I understood this correctly (from a mail from their mailinglist), the plan is to provide multi application support for DirectFB and then allow one process to be the XDirectFB server to load “legacy” applications. This really sounds like a great plan and the fact that they are rarely talking about it and much more coding makes me believe that they could actually reach this goal.
Another thing I just learned was, how the framebuffer actually works. The drivers are all compiled in and at boottime, it detects your card and loads the appropriate drivers (or VESA if nothing else fits).
Woooooooooo.
http://www.directfb.org/screenshots/XDirectFB-MultiApp.png
*drool*
Also, find increase font and decrease font should not all have magnifying glasses for icons.
Goodness yes. Magnify means look at something that isn’t changing sizes in the real world. Changing font size is about changing the physical size of it – zoom is a separate issue.
In some windows applications that I have seen they have a speed blur of a growing capital ‘A’ with an arrow pointing up and right. The shrink involves a speed blur with an arrow leading down and left.
That’s the kind of thing that deserves an animated icon.
I think OSnews should use slascode (or simmilar) no for the moderation thing, but for the sake of threaded coments.
PS. I posted once but don’t know what happened to my post.
I think Eugenia’s article really showed how much can be pared away off of KDE to get to the most intuitive form it can be. LOL, there’s a whole lot that can be pared away. For Joe User, the Lycoris Control Center is the best thing I’ve seen yet – and I don’t mean because it looks like XP, but because it is easy to use and friendly looking. You know, I wonder if it can ever be that an OS will have a system folder UI like Mac OS 9? I’m not talking about operating systems, but strictly UI. Even mid-level users of OS 9 have no fear of going into the System Folder after learning a few tricks. They know to go into the Control Panel Folder, the Extensions Folder and look for culprits there, know that a Preference, like the Finder Preferences might be corrupted and know to trash it and re-start. Just drag fonts to the System Folder to install them and it goes on and on. People can argue operating systems, of course, but I really do wonder if we’ll ever see anything that easy again.
There are few mentioning about disposing X. I totaly disagree with that since currently there are no viable alternative to it. Yeaaah, there are few project like PicoGUI, Microwindows, Berlin, OpenGUI, QtEmbedded, DirectFB etc but none of it reach a usable state for Joe user.
I’ve tried those and agree they are fast (especially PicoGUI and Microwindows) but lack of function except Qt Embedded give that more function but far behind compare to Xwindows. Xwindows also got other advance function that not availabe on others including Window$.
If any of you think X is the cause of the slow response, you better try EDE. No other DE in Linux world that very responsive and start faster (for me it start less that 1 second from the KDM login) whereas KDE itself spend more than 5 second. Is it really Xwindows fault or others?
I wholeheartedly agree with Eugenia – except for the “automatic disappearance” idea. IMHO there should be one basic rule of most basic rules: don’t outguess the user. Transparency is needed at every level. Unauthorized changes are one of the main reason Windows and other MS stuff irk me so much.
The HELP item may be unimportant for many users, but chances are those who start using KDE don’t use help for that first month you mentioned, Eugenia, and right in the second month they would like to take a peek at it – and it’s gone.
To be sure – I’m not a friend of those “Are you sure … ?” etc. requesters either that Windows bombards you with perpetually. Changes the user initiated do not need confirmation unless something vital is going to be affected. BeOS’s Tracker may be a good example again for how to handle that.
Qt Embedded works on FB and some KDE applications can already run without X (like Konqueror/Embedded and Korganizer/Embedded)
I love the authors constructive crtizism of KDE and as a GNOME user I would love to see a review of GNOME2 – bearing in mind that some bugs hinder usibility.
When I highlight and drag something, it should not copy it. It should move the highlighted text.
I hate this with a passion. It it FAR too easy to bork up a document (or worse, code) with an accidental drag and not even notice it until later.
When I drag a file to another directory it should just move it, it shouldn’t ask me if I want to copy it or move it. 90% of the time I just want to move it.
When I drag a file, regardless of the file manager I am using, I always right-click-drag to ensure that it asks if I want to move or copy.
I also wish they had spring loaded folders to make moving things even easier.
But I do agree with this
The reason I do not like KDE3.0 is because it is too configurable; this results in there being too many possible permutations to be tested.
I can understand why the KDE developers did this; they are UNIX users who enjoy the fact that X can be customized for every user’s taste. This result in having an option like allowing applications to either have graphical or text icons. Unfrotunatly, the person who wrote the game “Konquest” failed to test their application with text icons. As a result of this, “Konquest” has some of the window hidden when run with text icons.
Another option that KDE3 has which has not been well tested is the ability to remove the desktop icons. When this is done, and one changes the desktop wallpaper, one still sees the old wallpaper until one hides and reshows the portion of the wallpaper in question, usually by opening and closing an application with a large window.
Another bug is that active desktop borders are buggy; when first turned on, it can be difficiult to use the active desktop border underneath the kicker.
I also have had issues with how Kwin allows non-KDE applications to open up windows which are partially hidden by the kicker; in fact, I wrote a patch which resolves this issue [1] with my particular setup.
KDE is a very impressive piece of work. I find it very attractive, and deeply appreciate all of the hard work the developers have done on KDE. I hope, if a KDE developer reads this, that they do not mistake my comments as personal insults.
When people make applications for end users, it is important to listen to the UI guidelines that UI experts have. Someone who knows C (like myself) is usually not a UI expert (like Eugenia).
– Sam (http://www.samiam.org)
[1] http://www.samiam.org/screenshots/kde3/kdebase-3.0-kwin.patch
Concering usability of X11 for remote use: yes, right now it’s mostly for sysadmins. But even a feature only for sysadmins can be useful… beside that, all that’s lacking currently is the right frontend for end-users. There will be one in KDE, but in KDE 3.2 or later.
Attempts have been made, but they all remain half-baked. Wishing for a replacement is easy, but writing one, and actually getting it to the point where it is useable, (and able to support existing X apps), is an incredible amount of work. I was hoping that KDE would actually create that replacement eventually, but judging from past interviews, that possiblity would seem to be remote.
I don’t think being usable means running X applications. What is the use of bring inconsitency into the GUI again? It is not like there is such important applications Joe Users use that aren’t available else where. Running X applications shouldn’t at all be a goal, you would be just another extended X implementation (remember Metro? The offered some really cool stuff but nobody wrote for it).
And the only way for these “half baked” graphics servers are GNOME and KDE supporting it. By porting their desktops there, making their apps consistent and so on would do great for it.
Turn on autohide if KDE is capable of doing it. Then you’ll be taking up NO space for the taskbar, but when you need it, it pops back up with all the room you need and then goes away
Defaults remember? Eugenia made a point that not much people changes the defaults.
There is no reason that a system like Mac OS X couldn’t be implemented on top of X11. You could implement Display PostScript (like with GNUstep) or Display PDD (OS X’s quartz) on top of X11.
Actually, no. Mac OS X’s Aqua couldn’t have possibly be implemented on X. Firstly, Quartz is a vector windowing system, as oppose to X’s raster self (though a lot of raster-based stuff like the icons are there). Secondly, a lot of features OS X had implemented couldn’t had possibly worked on X11’s plain vanilla specifications, for example Unicode (X11r6 doesn’t support it) or true alpha blending (X only recently got it with copying the background of the object and blending it in, quite slow).
Organizing a big push to get all the major distrobution organizations to start using a non-standard X server would be a lot easier (and more practical for all involved parties) than would to have it Berlin (or something akin to it) replace X11.
Believe it or not, Fresco/Berlin is not here to replace X11. Heck, before 1998, Berlin was written in assembler, and Fresco for a X11 toolkit. Fresco differs a lot from X11, like it is being a pure vector window system, as oppose to being a raster system. Sounds good? Not really, with current hardware technology, it is slower.
We have to understand that Unix isn’t just _one_ desktop environment (and now please don’t respond with an upset “that’s why Unix will never make it to the desktop”) but many. Just like we have Windows, Mac OS and Amiga OS, we have KDE, Gnome, XFCE, etc.
Notice on Windows, you could switch the entire desktop to say, LiteStep or Object Desktop, and all applications that properly supports the Windows UI instead of cloning it in their applications can blend in. What Eugenia wants is that to happen on Linux.
Even Microsoft isn’t consistent in its interfaces: look at the difference in menus, toolbars, dialog boxes, etc. between Win95, Win98, Win2k, WinXP, Office 97, Office 2k, Office XP, etc. That’s not counting other applications…
That is one of the problems of Office, it doesn’t use Windows’ UI.
Personally I think the KDE interface is horrendous.
Counter-flame: It is, but I think GNOME’s is worse.
(Haha, I just use KDE cause it is the one that suites me best. I like Window Maker’s look best).
The article states that normal users should not be allowed to use the quick browser for the / directory. Why? This would be just plain annoying. If they are allowed to acces a directory, then they are allowed to access a directory. If they aren’t, they still aren’t. Try browsing to some directory you are locked out of, and you get Failed to read directory, which works.
For normal Joe User (why Joe? Who’s Joe anyway?), i don’t think / is a good thing to show. Sure, for geeks like us, it is hell if we were blocked from it, but from them, they would wonder what the heck is /usr or /dev (never go there on a Mandrake machine with nautilus 2..)
Linus has *no* work to do about nVidia drivers – I have a nVidia card on one machine, and it’s a PITA. Definitely not a Linus/Linux problem, though, just a p*ss-poor driver from a 3rd party. That’s comparable to saying, “Windows sucks, it crashes whenever I plug in <some specific hw item>”… 3rd party driver problems are not an issue for the UI or OS teams. nVidia just need to catch up.
No, the problem is with the Linux kernel. It breaks binary compatible so often. NVIDIA has an policy of non-open source drivers, which makes it hard for users of newer kernel to use NVIDIA cards.
On the software side, the Gnome guys are porting the Gnome/GTK framework to DirectFB. No word on KDE though. Maybe it would be a good idea to start on a unified desktop for running under DirectFB. Are the KDE and Gnome developers reading this!!!
KDE would be ported there as soon as there is a stable Qt implementation there, under a free Software license. And until now, there isn’t such a thing. Or perhaps a toolkit better than Qt that does C++ the way Qt does and makes porting easier from Qt to that.
…companies like Xandros and United Linux take on.
The last I checked, UL is for the enterprise… I see no need for KDE there…
Your suggestion are quite good, but since as you say gnome2 is what you feel need more improvement, please make an suggestion article for it too ^^
Great article on the UI, but I disagree with the proposed approach to integration:
You can’t and shouldn’t just use DrakConf, Yast2, or apt-get, or portage. These are just tools for four distros, tools incompatible between them! There are more than 10 important distributions today! What about these rest of distros? There has to be found a unified way, a way that works universally to all these distros, a way endorsed via the desktop environment.
Reader comments:
To me the best solution would be for kde to have a package installer that does the ./configure; make; checkinstall; (replace checkinstall with something for your distro’s package system) for the user. The user would not have to see the details
The trouble of compiling something isn’t typing “./configure; make; make install”, it’s figuring out why it doesn’t work.
1. Ordinary users will install and remove software as packages. What they need is a good GUI tool for managing packages. rpmdrake and kpackage are good; possibly some users would benefit from a system where they can opt not to see most of the huge number of packages on the system, unless they cause a dependency problem, for example.
2. Ordinary users should NEVER, EVER, EVER, need to build software using ./configure, make, make install: at worst, they should build from a source package, via the GUI front end to the package manager. The most common reasons why ordinary users need to build from source tarballs are:
(a) the software is only provided in that format;
(b) the software is provided as an RPM, but the dependency mechanism does not work because different distributions use different names for the same packages.
We need to tackle these two problems, not try to create a workaround so that non-technical users can attempt to build software, without even knowing whether the correct libraries are present, or what to do if there is an error.
This articles points are well taken, and many of them I agree with. As someone who has worked on KDE usability, via mailing lists, it rather frustrating to read yet another laundry list of criticisms (however well reasoned) of KDE without many answers on how to fix them and how to recruit someone to get the fixes made. KDE has a forum for this kind of work: kde-usability at kde.org. Please consider joining this list. Otherwise your points will remain academic until someone decides to do something about them.
You probably see pretty quickly on the list that a) things are kicked around for a while, as not everyone agrees, and b) it’s usually up to the person with the idea to recruit a developer (or code their idea themselves).
Cheers,
Eric
Of course, the irony in all of this is that Gnome and KDE, Windows clones, are progressing more every day, while the truly revolutionary graphics engine Fresco (formerly Berlin) is moving slowly. With Fresco, the developers had the bright idea that a graphics engine could be made that supported both network transparency and easy resolution-switching; and also transparency, and the whole thing works at a much higher level.
All the developers working on imlib, gdk, gtk, glib, etc. should stop immediately, and begin working on Fresco. Then, all developers from either KDE, Gnome, or other desktop environments on non-graphics parts of the libraries (such as drag-and-drop) should join into a single group under the domain of FreeDesktop.
With luck, 6 months from now we’d be able to have Fresco 1.0. No longer would Gnome and KDE be separated by different languages and toolkits. You could finally use konqueror and gnome-panel interchangeably if you wanted, and they wouldn’t look any different. You’d only need one set of libs.
Oh, why can’t anyone but the Fresco and OBOS guys have any foresight… do they lose it once their project gains significance? Linux used to be innovative, and now it’s just hanging onto old standards and copying Windows. Let’s hope BlueEyedOS fails before it’s ever released.
I already have. I joined the discussion yesterday.
1. “Mounting” a disk or other media needs a more friendly word. No one but a Unix-head has ever heard of mounting a disk… as it hails from old tape drives (?)
“Detect” (for mounting) or “Remove” (for unmount) might be a superior starting point.
2. “Kblah”, “KBlatz”, “KBlooie”.. get OFF it. The K prefix promotes a weak core brand notion at the expense of legible, clear names for applications and utilities. Toss it as a naming quirk.
tone
I hope you submitted your comments to the KDE usability team as well. I think you have a lot of good ideas about KDE and Linux in general, but it might appear that you’re doing much asking and not much doing. Of course I have no idea if that is true or not, but it would be good to mention in the article or in the feedback here if you’ve done anything to help out with the problems you see.
Both the KDE-devel-Core and KDE-Usability mailing list people are aware of the article. They are discussing it, as we speak.
Very nice article!
I have a comment on the Qt/GTK collaboration idea though:
I think it would be easier if the KDE people wrote a Qt/KDE Style that mimicks GTK appearance than writing a GTK theme that looks like KDE.
Same for the other side.
And actually I think it should be possible to make a Qt style that looks almost exactly like GKT.
Qt was designed to look like another GUI.
Cheers,
Kevin
> it would be easier if the KDE people wrote a Qt/KDE Style that mimicks GTK appearance than writing a GTK theme that looks like KDE.
This does not make sense at all. Think a bit. Why would someone want to mimick GTK+ on a KDE desktop? What we are trying to achieve here is consistency, consistency with ones desktop, not with the other’s.
Ok, for one thing, UNIX WASN’T MADE FOR INTEGRATION!!!!
Now that I got that out of my system, here’s some real info for you. If you want to install new software w/o using cmmi, just use Mandrake. Its urpmi works wonders for that kind of stuff. And you have confused “Device driver” and “software” so much that I wonder if you really are a *NIX user. Device drivers are KERNEL MODULES! These can’t be installed with just a simple front-end installer, it’s just plain stupid. If you want to administer a system you must understand the system. And a PBN Sysadmin (That’s “Paint By Number System administration” for you to comprehend) is NOT system administration. Therefore, if installing device drivers is a sysadmin’s task, then the sysadmin should know what’s going on. Instead of just pointing and clicking ‘install this thing that I’m too stupid to know how to install’, you just run a VERY SIMPLE command line (I’ll show you, since you don’t seem to have the IQ to understand it):
modprobe <module_name>
where <module_name> is the name of the device driver module name. Now, beware, that is a linux command, so it won’t work on any other OS.
Oh, and by the way, integrating the GUI with a Unix kernel is the WORSE possible thing you could think of. The whole point of ANY unix is to be command-line based. Now, I know that Apple did it with the MacOSX, but even they layered it, so that it is a microkernel, and therefore can decrease full system crashes. As a note, take a look at the earlier MacOS’s and Windows{NT,2k,XP}….they experience crashes that destroy the entire OS. This is NOT something you want with a Unix OS.
Ok, enough rant, time for lunch.
Justin Hibbits
> Why would someone want to mimick GTK+ on a KDE desktop?
Let me rephrase it.
If the KDR developers made a Style that looks like GTk, KDE apps could use this style when running under a GTK desktop and thus look like GTK apps then.
I just wanted to point out that it might be easier for KDE developers to write such a KDE style than it would be for GTK developers.
The latter was my interpreteation fo what you wrote in your article.
Cheers,
Kevin
Ah, now it does make more sense.
However, we never know if the KDE guys write such a style if it will ever be used by the Gnome 2.x guys (you see, this team will need to work to tell the KDE apps to load a specific theme). And of course, the opposite also applies for Gnome apps on KDE. The KDE guys need to make sure that the do load the right theme.
And with your idea, there is antoher problem. Not all the people will have the new KDE or Gnome version that includes the GTK+ and Qt themes. So, it is better for the KDE team to write the GTK+ theme and load it on demand, and the Gnome team to write the Qt theme, than the opposite!
BTW, I tried to email you earlier, you address does not work.
> this team will need to work to tell the KDE apps to load a specific theme
Right. I think every KDE app has a commandline option –style that tells it witch style to use.
Perhaps this could be automated, again by KDE developers, if KApplication checked on startup if it is running on, lets say GNOME, and apply the GTK style then.
> Not all the people will have the new KDE or Gnome version that includes the GTK+ and Qt themes
Ah, I see. I also think the other project should distribute it.
I just think that writing a KDE Style is far easier for a developer who has KDE experience.
And those are more leikely be found in the KDE developer group.
Same for the GTK side.
Well, that is if GTK has something like the style engine of Qt.
(Qt styles are compiled plugins)
> BTW, I tried to email you earlier, you address does not work.
I am sorry, I didn’t know that the account got closed.
HAve changed to a working address now.
Cheers,
Kevin
So you are saying that Unix based/like systems are no place for the average non-sysadmin user that needs to maintain thier own system (ie: not at work where they have an IT dept. for such things). Fine, see where that gets Linux on the desktop.
As for modprobe, why couldn’t you have a ‘Module folder’ that, when a module is dropped in it, will automagically call modprobe <module_name>? As a bonus, it would also show loaded modules. As a bonus bonus it could show more friendly info than the typical module name. If that sounds too much like the classic Mac OS system folder, well … it is supposed to. They were on to something there with regards to simplifying system maintenance tasks.
Why don’t I ever hear anybody about KPager. I WANT PREVIEWS IN MY KICKER! I don’t want to click a button and then see an extra window with my desktops. I want to see them right in my kicker, previewed and all. That’s what I loved of Gnome’s panel… dock… I forgot what it was called, but I loved it.
What you are suggesting is a change in the Linux kernel. In order to support the ‘system folder’, you have to patch all the modules to give a detailed description of what they do. I’m probably one of the few people who actually believe that Linux shouldn’t take over the desktop. My reasons are that Unix is Unix. Therefore since Linux is a Unix clone, Linux is Unix. And if you think back to the design of Unix, user-friendliness was not their top priority, it was getting work done. And going through a point-and-click interface is NOT getting work done, it’s wasting time. The goal of a system administrator is to get the system up and running, and to keep it running. Therefore, he must be able to quickly load and unload modules (hence modprobe), which navigating the directory tree to find the ‘system folder’ is a waste of time. I can load a module 10x faster through a command line than I ever could with the POS GUI module loader/configurator that comes with the distros (forgot which one, prolly redhat).
As another argument, non-sysadmins wouldn’t be installing hardware, they would take their system to a ‘qualified technician’ to get it installed. Either that, or they would (like my sister) ask a geek relative to do it. but either way, adding device drivers is not the job of someone who doesn’t know what he/she is doing.
I do have a feeling that I’ll get flamed for this, but I really don’t care. These are my views. Linux wasn’t designed to be used by someone who only knows how to bootup and use IE to browse for pr0n. It was designed as a pet project for this guy who just wanted a non-M$ command line to mess around with.
Maybe I would be motivated to write a Gtk-style for Qt if I were using any Qt application but as long as that’s not the case, this is not gonna happen. =) The main reason I try to avoid Qt btw is not about different looks (I couldn’t care less) but because those are slow and take some extra memory when running in a Gtk environment, especially KDE applications (Konqueror loads slower (10 seconds) than Mozilla (3 seconds), on first launch that is).
I’m pretty sure that this is the reason why the idea of integration comes up often, but then gets forgotten when people really investigate into the problem. IMO the better idea is to seperate those systems more and let them grow each on their own (while keeping functional compatibility).
I think as soon as KDE has something like Gimp (Gimp2 will be a bit more toolkit independent, maybe someone writes a KDE frontend), both desktops should have everything a typical user needs.
That is bullshit, making a hardware work is not the job of a so-called “geek” or a “qualified technician”.
The problem isn’t modprobe. Distributions can autodetect hardware and then modprobe them without a problem. The “discover” application can do this even silently at boot time.
The problem is only when installing a _new_ driver. You can’t just download the binary and place it into your /lib/modules folder because there is no such binary compatibility. You have to compile the module against your kernel and this depends that you actually compiled your kernel. Distributions can offer such modules compiled against a specific version of their kernel but it’s really a mess.
You say that Unix wasn’t done with userfriendlieness in mind, but don’t forget that we are using GNU (and GNU is _not_ Unix as we all know
) and GNU actually has userfriendlieness as one of their top priorities.
Hrm…I’m using GNU? I didn’t know that….Lesse….compiler is GNU, debugger is GNU, binutils are GNU…sh-compatible shell is pdksh, login shell is tcsh, editor is vim, GUI is X w/ windowmaker (GPL, but NOT GNU), mail client is mutt, sed is ssed, tar is star. Most of my base software is NOT GNU. it’s either BSD or my own, or other software. As you can see, my system can’t be GNU (oh, I am using glibc until I can get a viable alternative to compile and work properly…had problems with uClibc).
as for the Bullshit you’re speaking of, I’m coming from experience. The only people who install hardware around me are either me or someone else who’s competent with computers, NOT the Joe IELuser who just uses the computer to surf the web, play games, do taxes, and write some wordprocessor documents. Now, your experience may be quite different, but I’m going off of experience.
And you can continue to use GNU, I’m gonna continue to use my hacked-up Linux-based system, while I continue working on my Gold project ( http://www.gold-project.org ) to completely eliminate all GNU stuff.
Justin Hibbits
TLy : “Would it be too much to ask for a Linux distro made specifically for KDE so that it may be optimized for KDE and really integrate KDE into the underlying system?”
That’s exactly what I’m working on!
It’s basically RH 7.3, with 100% Qt/KDE, a slightly simplified GUI (I will incorporate some of the comments from this article), more KDE integration with the underlying OS (this is what still needs the most work), and a perfectionist’s attention to detail.
Splitting the non-Qt stuff out of RH 7.3 was a major undertaking by itself.
I (perhaps facetiously, perhaps just hopefully) started calling it “Project: Real Choice”. That got shortened to “Perc”, as in “Perc Linux”.
I also think clean APIs are important. I have been impressed by some APIs (Qt, BeOS, even libowfat), and utterly disgusted by others (libc / ancient *NIX). And I have also been disgusted by the lack of APIs in certain areas of Linux (run levels, disk detection, the /proc interface being treated as an API, …) Ideally, as time goes on, I’d like to clean some of this up and make a tighter system.
In case you can’t tell, I love Linux, yet hate it. And that’s what’s driving this project for me.
I’ve got http://www.percos.org registered, but no server for it yet.
I have SRPMS, docs, rants, TODO lists, and more. I’ve got more ideas than I have time.
I am so used to hearing fellow Linux people insist that it’s the users who are stupid, rather than Linux that is stupid, that I haven’t shared this with anyone. For now it’s just for my computers… but if anyone actually is interested, let me know. We could at least start by sharing docs/ideas/TODOs…
Personally I can live with a badly designed and limited UI more easily than I can live with a mess of different apps that all work in different ways. In a lot of ways KDE has a better designed and more powerful UI than Windows. But until there are KDE apps that do everything I want I have to mix in apps created with other toolkits. Even the real basics like cut/copy/paste and drag&drop are often totally different and incompatible between different apps. Windows has it’s annoyances, but there’s nothing even close to being as bad as that in it’s GUI.
Maybe someday KOffice, Konqueror and other KDE apps can replace OpenOffice, Opera, Evolution, GIMP, etc. and I can have a consistent GUI in Linux. But until then I don’t consider it to be as good a desktop OS as Windows despite the Linux advantages.
I wasn’t talking about you specifically but what we are talking about here. If you prefer to use something different, fine, go on! Your language and intolerance against people who prefer something different than you really fits with your project pages description.
Pace Mr. Hibbits, out here in the real world ordinary people often must install hardware, without the benefit of friendly geeks. I know I’ve had to, and if it wasn’t for plug and play, I wouldn’t have been able to. And installing new hardware is exactly the sort of thing an ordinary user should be able to do without calling for a service technician. The entire purpose of a desktop computer is to allow ordinary people to do more without technical knowledge. Real power is power you can use, or else it’s meaningless. You Linux guys are arguing over ease-of-use issues that were settled a long time ago, as far as Microsoft and Apple are concerned. If the Linux community really wants Linux on the desktop, the whole community — developers and users alike — must get over their conceit of themselves as an elite.
I was just wondering, where could we find the modified .NET theme you mentioned in the article? Thanks.
On my hard drive… 😉
Or, you could edit the source yourself and recompile it.
If you can look past the flames and smoke coming out of your eyes, you might see a group of people right here as well as everywhere, who are trying to make their favorite OS better. The computer industry is not run by sys admins. It’s the little people that buy products and services and pays the big companies that pay your DAMN salary.
I don’t particularly care what you think Unix was made to be like. This is not about what Unix is, this is about what Unix will become. You can’t expect it to be in command line forever. All this effort to make a user friendly desktop with Unix underneath is to prolong the life of Unix, not change it and destroy it.
As far as stability goes, we’re not suggesting the merge of UI code into the kernel thus increasing the chances of system wide crashes as you have described. Instead, we’re suggesting a more standardized way of doing things, a way that is easier to use and not have to carry with it the burdens of Unix.
Apple had the right idea of using a BSD microkernel wrapped around a Mac shell. When I first mentioned integrating KDE into Linux I was thinking of exactly this. Pretend it was the Linux kernel instead of FreeBSD, XFree instead of Quartz, and KDE instead of Aqua. That is what I want to see.
What I would love to see is something like XP’s active context elements when using browsing your hard disk. Having a context menu at a RMB click is nice, but it is so much nicer to have context-aware options just one click away and VISIBLE on the screen. Same goes for searching…
This could be integrated in the sidebar but could also be placed directly inside the browsing window (see XP).
I know some people will hate this idea, but despite me being an absolute fan of KDE, when I first worked with XP that feature really hit me as being great for both beginners and experienced users.
Why are KDE/GNOME menus so wide? They take so much space unlike the Windows ones. For example, the right click menu for a text box, or the desktop, just see how wide it is.
Look at the Run Command diolog in the following screen shot, the context menu for the Edit box, just look how wide it is, it takes so much space.
http://static.kdenews.org/mirrors/qwertz/kde31alpha/2329-1.png
That is common with all menus and dialogs. Things which in Windows can be fit in a very nice small dialog box, in KDE, the same dialog box will be almost as big as the screen in a 800×600 resolution.
Well, i’ve been waitng for the 100th post to say this. I thought the article had alot of good ideas, and hopfully the KDE people will implement some of the ideas in the article. I especially liked the part about kicker, which i’ve always thought could use a lot of work.