Computerworld: KDE 4, the Ultimate Business Desktop

Australian computer news site Computerworld asks if KDE 4 will be the ultimate business desktop. Speaking to developer Hamish Rodda they look at the changes being made to the KDE libraries including the Akonadi storage manager for PIM data. He also explains why KDE 4 will be important for ISVs to support.


Yup, on my openSUSE 10.2 box as well :)

And I think font-issues should be handled bij X or the distribution, not by the desktop environment (since not all applications use the desktop environment)

By otherAC at Sun, 2006/12/17 - 6:00am

> I think font-issues should be handled bij X or the distribution, not by the desktop environment

but KDE developers have fixed quite some issues with the freetype lib (that handles fonts)... i even read some KDE dev are know as the "performance guys" to the freetype devs.

By cies breijs at Sun, 2006/12/17 - 6:00am

yes, and they have fixed it in the original freetype lib, which is used by all graphical stuff, so all the fixes are not just KDE-specific.

By Jakob Petsovits at Sun, 2006/12/17 - 6:00am

Care to post a screen-shot? Thanks.

By cb at Sun, 2006/12/17 - 6:00am

I tried to read the article a couple of times, but it looks like the site is down. Anyone knows what is going on?

By Tomas at Sun, 2006/12/17 - 6:00am

probably to much traffic from the dot to the server :o)

By otherAC at Sun, 2006/12/17 - 6:00am

i think the article makes a good point - management. i think this is an area in which linux really is behind windows. i work in a company mixing linux with windows, and if we on linux have to make an account (NIS) we have to edit passwd, shadow, etc by hand. giving a person permissions on a folder? a lot of work.

the windows guys just point'n'click, and can even set a time the user has to be removed automatically. they're 10 times faster on the basics.

ok, we could write scripts. but the company (as any company in the current fast moving world) changes a lot, so the scripts have to be rewritten every - what, week? business units change to divisions, they want their projects indexed by year, but a year later it should be the newly ("we're back") business units, etc.

ok, linux is more powerfull. complex things can sometimes be done much more easily on linux. great. but in the basics, windows RULES. and what constitutes of 'the basics' is always expanding, thanks to the hard work microsoft and many other companies delivering system management tools are putting into it.

linux needs easy to use administration gui's. yeah, the commandline is powerfull, more efficient compared to a gui. well, forget it, that was years ago. it might be the case for linux gui's, as those are mostly a very one-to-one mapping of the commandline stuff. not for windows, tough.

i hate to see those windows guys do much more administration much faster than we linux guys, and i think that's something which needs fixing...

we have the basics worked out here, in KDE. i think a current KDE desktop kicks windows XP's ass, and i hope around KDE 4.1, it'll kick Vista's ass. but a company needs a bit more than a flashy gui, it needs to deploy systems, manage users, set availability of software, etc - and not by writing (and rewriting) scripts every week...

By superstoned at Sun, 2006/12/17 - 6:00am

I think you should have a closer look at the current state of linux configuration :o)
lots of stuff you mention is also point and click in linux.

By otherAC at Sun, 2006/12/17 - 6:00am

Hmm, since you are in a mixed workstation setup, wouldn't it be easier to use pam_smb instead if NIS to administrate the user accounts?

By Kevin Krammer at Sun, 2006/12/17 - 6:00am

I read in my Linux Administrator book that NIS is really still only in Linux for backwards compatibility, and is especially not good enough for mixed Windows/Linux environments (since NIS is UNIX-only). Yes, you should probably look into pam_smb or pam_ldap...

By Dark Phoenix at Sun, 2006/12/17 - 6:00am

You really need to redesign your authentication infrastructure if you're editing passwd files by hand. Jeez... that's archaic. Use an LDAP server. If your network has an Active Directory server, you already have one (although somewhat bastardized by ms). Then, use plugins for PAM and NSS (usually packaged as libpam-ldap and libnss-ldap. Also look for a pam plugin that automatically creates users home directories on their first successful login.
When an account is created on the server, the accounts just work on all the workstations. If you want to have users home directories available from both windows and linux systems, and you're using a Samba server, make the home directories available to the local network via NFS. Since the user database is shared among all the workstations, user numbers will match everywhere and, again, it will just work. This is far from the only, and probably not even the best, solution. However, it does show that you can do far, far better then editing files by hand to add users.

By Benjamin Long at Tue, 2006/12/26 - 6:00am

Hopefully, KDE4 will be smoother in Desktop redrawing then KDE3 and runs on a 500Mhz as good as a window manager and/or Gnome.

Maybe KDE4 will be better from the technical point of view then *every* other Desktop. But users don't want to see sloppy Desktop redrawing because this isn't ergonomic.

By Anonymous at Sun, 2006/12/17 - 6:00am

What do desktop redrawings have to do with ergonomics?

Anyway, kde4 is a lot faster then kde3, so chances are that redrawings wil go smoothly on your system.

Choosing a lighter linux distribution might als help (dunno, which one you use now)

By otherAC at Sun, 2006/12/17 - 6:00am

KDE is not a window manager, it's a Desktop Environment. If you compare it to Gnome on your machine, what figures can you give us about KDE 3 not being up to Gnome in 'Desktop redrawing'?

By Anonymous at Sun, 2006/12/17 - 6:00am

Everyone who wants to see KDE4 screenshots, here you are:

By Lans at Sun, 2006/12/17 - 6:00am

The observant reader will notice that it looks much like KDE 3, so before you give yourself indigestion over that, remember that software is like an iceberg - 90% is below the surface. Anyone remember the same deal looking at the KDE 1.8x versions?

The past 18 months have been spent in porting and rewriting the APIs that make those apps possible: agreeing a style standard, tidying up code, documenting, making interfaces cleaner, throwing out wholesale bad API, duplicate functionality and functions that turned out to be only used by one app. We know what worked and what didn't after five years of being bound to the design decisions of KDE 3. As an aid to ourselves, and as a gift to the new contributors who will join the project in the next five years, having this extended basic maintenance period for the platform is essential.

This refit is now mostly done, and developers' attention can move to the user visible parts, like applying usability experts' advice (and 5 years of dot comments), writing nifty new widget themes, and writing cool new apps with greater ease than before - and major distros (openSUSE, kubuntu) are providing preview packages of the libraries to speed up porting.

By Will Stephenson at Sun, 2006/12/17 - 6:00am

Yes, the widget drawing will be smoother, thanks to Qt4's Arthur. And you won't even need an expensive proprietary driver to get the benefits of it.

By Brandybuck at Sun, 2006/12/17 - 6:00am

Thanks! One great answer.


By Anonymous at Sun, 2006/12/17 - 6:00am

proprietaty... ok...
but what is the last time you paid for the driver :S

By Mark Hannessen at Mon, 2006/12/18 - 6:00am

There is an interesting comment on the LWN entry for this article
( ). For convenience:

The way window managers work (user interface side, not code) hasn't really changed since the start of X windows, many eons ago. Amongst the plethora of choices, they differ only superficially: there is still a titlebar as well as the close/minimise/maximise buttons and the ability to resize windows.

How about something along the lines of a window manager which enforces a particular structure and uses the available space considerably more efficiently? e.g. a "tabbed" window manager, ala Firefox (this is more than just having a taskbar).

An example: windows would not be allowed to be moved and resized arbitrarily and the annoying pop-ups (e.g.. file-requesters, error messages) would not simply (and rudely!) cover other windows. If one has a browser/editor that takes up the entire screen, and then launches a terminal (or some other window wants to open up), the size of browser/edtior would be reduced by half and the terminal "window" would take the bottom half (or the right half) of the screen. There would be no titlebars - on a laptop, where screen space is premium, having multiple title bars amounts to wasting space. Actions for closing the windows would be done on the tabs.

By Mark Tall at Tue, 2006/12/19 - 6:00am

I think it's a good idea that you would check out the many ways you can manipulate the way windows behave in KDE.

Most of what you suggest has been available for ages in Linux desktops / window managers.

By otherAC at Wed, 2006/12/20 - 6:00am

The window manager in KDE does not enforce the originally described constraints in a systematic fashion. It can have specific settings for specific windows which go partially towards a window manager with "constraint policies".

Kwin is still an old-style window manager that just happens to have been written relatively recently. I'm not saying it's a bad old-style window manager, I'm saying that it would be handy to move beyond the current windowing paradigm.

Kwin doesn't, for example, automatically re-size the dominant window (e.g. a browser) when a new one opens up (e.g. a terminal). It doesn't have an option to enforce a policy of "disallowing any kind of window overlapping".

There are perhaps only two new window managers (ion3 and wmii, which stands for window manager improved 2), that (mostly) implement the described contraints. However, they do not fit in too nicely with the rest of KDE (e.g. panels).

By Mark Tall at Fri, 2006/12/22 - 6:00am

Yeah, and that's particular important now that screens are getting bigger. In fact, reading a webpage in a fullscreen browser is not pleasent in modern wide screens, and resizing windows manually dragging your mouse is not pleasent either.
I'm not sure if the solution is revamp automatation though. I think that when you press the maximize button, it should go to a sensitive size, and you'd press it again to maximize more or fully. Its funny that people use more than one screen to be more productive with the desktop management, because then they can just press maximize and have the window fit nicely.

Unfortunately, I think proprietary systems, like Macs, will get this first and free software will just play catchup like in most user interface elements areas.

By blacksheep at Thu, 2006/12/21 - 6:00am

Although I'm not a developer I can imagine that it is a lot of work to release a new desktop environment. I think the developers are doing a good job when it comes to programming.

But I think there are some other problems in the KDE-project, especially in KDE4.

1) Project management

A lot of people complain about Microsoft and there release-delays. But on the other side, a lot of open source project do exactly the same thing (Debian, Mozilla, KDE and others). If you are in a position like Microsoft it might not be a big problem, because a lot of people have to use their OS, and a lot of people don't even know about other operating systems.

But if you are in a position like KDE where you are trying to gain market share, you sould not repeat their faults, you should try to do better. There is not even a timeline at the moment, at least as far as I know. And once again, if you are not in a position of having about more than 90% of market share the "it will be released when it is finished" is not very helpfull.

And this might be one reason distributions like RedHat and Novell seem to prefer Gnome at the moment, just because they know what the will get in the next release and when they will get it. This is NO "Gnome is better than KDE" or "KDE is better that Gnome", I use both of them - some things I like in Gnome, some things I like in KDE. Maybe Gnome will fall far behind KDE4, but who really knows?

No timeline, no milestones, almost no information about the progress, no information about remaing tasks and their needed time. Don't get me wrong, but from outside of the project, this does not look very professional.

2) Information management

There are a lot of websites, spreat all over the Web:

KDE4 goals (
Release Plan (
Appeal (
Plasma (
Solid (
Phonon (
KDE-radio (
Oxygen icons (
Decibel (
Wikipedia (

The problem is, most of them are not very usefull to users, because there are only very vew information on this sites and even less news. Most of this websites are completly out of date.

Why not one Website related to KDE4, let's call it for example. This site should/could contain all the related stuff as subsites, like (main infos about the project KDE4, what it is, etc.) (regular updated (!) site with all the news) (detailed goals for KDE4) (or releaseplan, whatever you wanna call it) (infos to this subproject) (infos to this subproject) (infos to this subproject) (infos to this subproject) (infos to this subproject) (all documentation stuff) (when available) (snapshots including needed infos how to build, compile, install...)

This website could be one site (including subsites) for both - users and developers. I think this could satisfy users in being informed about the status and the development of KDE4 and it could help developers who want to start developing KDE4 apps. One important thing is update, as mentioned above the complete site should (almost) be up to date all the time.

I really hope you guys don't get me wrong, just a few points I have noticed in some critics about KDE4 and some ideas.


By Ralf Weyer at Fri, 2006/12/22 - 6:00am

> And this might be one reason distributions like RedHat and Novell seem to prefer Gnome at the moment, just because they know what the will get in the next release and when they will get it.

Since the KDE 3.5 release plan shows a expected date for the release of 3.5.6, I assume you are referring to KDE4.

However, I am not sure if GNOME has even started with GNOME3.0 development, so I would be surprised if they had already a release date for it.

Or it could be a misunderstanding. Do you mean that KDE's 3.x released were not as well scheduled as GNOME's 2.x releases?

By Kevin Krammer at Fri, 2006/12/22 - 6:00am

It is not about any number or special version. Numbers do not count that much I think. It is just about the latest version of the desktop environments, so Gnome still the 2.x-series and KDE it is version 4.

When talking about Gnome, at this point there is not planing of version 3 (only so very undistinct thoughts about it).

But I don't think a major version should be handled different as a minor version. You allways have to plan it, you allways have to have a good project management, you allways have to have milestones, ... It is just about how to handle such a project, and the bigger the project, the better the project management has to be.

And information to the world outside of this project is very important, especially in open source projects. This could be one of the biggest advantages of open sources software comparing to propriatary software.


By Ralf Weyer at Sat, 2006/12/23 - 6:00am

This is a major misconception from a non-developer. Consider the fact that over the years, KDE has built up an extensive amount of code. Additionally, consider that changes to the API were disallowed during the KDE series of releases to allow commercial ISVs to have a stable API to work with. Now look at the preceding with these nuggets of thought. KDE 4 will be using Qt4, which has a wildly different API than that of Qt3 meaning lots of changes that MUST occur. Additionally, the kdelibs developers are finally getting the opportunity to clean up the code that couldn't be changed during the 3.x series due to the API guarantee. We all just have to be patient with this, think of it this way: when KDE 2 was in development, it took a sizable amount of time to get the code changes done too. GNOME had the same issues when they were moving from GNOME 1.x to GNOME 2. Additionally the Linux Kernel has had its share of growing pains when moving from the 2.4.x series to 2.6.x. This is hardly a KDE project mis-management issue, but rather a normal occurrence when you have a sizable code base such as KDE.

By Gary Greene at Sat, 2006/12/30 - 6:00am