The roads I take...

As I mentioned, I had planned to switch to KDE 4.2 on my production system as well after it worked pretty well on my laptop. Additionally, most people in our communities probably know about the value of testing development code in production or almost-production environments to find all kinds of bugs and get them reported upstream.
And so, I finally did update my openSUSE Factory system again - it's been a number of weeks since the last update, and that's probably not ideal in the case of such development distributions (but the state I had so long was fairly near to the 11.1 release).

The first problem came up right during the update/install process: package files were pulled out under my feet while I still needed to install them. I guess there is some room for improvement in the otherwise pretty good openSUSE Build Service (OBS) used to provide all those packages. After a restart of the install and another install/update run for more such problems with another OBS repo I've included, I finally had all the new packages on my system. I also had cared to replace the KDE 3 login screen with the KDE 4 one and install all packages I should need with the new desktop after the switch. So far, so good - it was about 4am and I decided to go to bed and continue today.

So, after some hours of sleep, I went up, not forgetting that I still would need to re"compile" the proprietary nVidia driver package (actually, it recompiles the kernel interfaces to load its binary blob with and installs all that). I couldn't help but notice my harddisk churning like mad though and realized an "old friend" from the last such update had come back - the logs were again filled with ATA errors, and writing a few of those every second to the on-disk log really slows down the system a lot. A few kernel reinstalls (trying different versions) and bug comment entries later, I found the only way I could stop this for now is by shutting down the syslog service and not logging any such info on disk for now. At least the engineer owning this bug should have more info now - that's what using development code is all about, right?

So, with the logging off and the video driver installed, I finally logged into KDE 4.2 (or actually, because I'm livin' on the edge, 4.2.60, i.e. KDE "trunk", which is already on the way to 4.3) and started configuring it. I knew most things I needed/wanted from the laptop already, but I mainly ran into two issues: 1) Those nice semi-transparent effects wouldn't work and 2) I wanted different background images for my 8 virtual desktops but could only set one for all.
The first problem was explained in the settings panel with a missing XComposite extension and I could find out that the nVidia driver had set "Composite" "off" in the config file while editing that to "on" just made this work.
The second problem required some more searching, but I could find that the backend work was done as well as an option to turn it on manually - which worked far enough that the background I had set is now on only one virtual desktop, but all others are black and empty (i.e. no background, no icons, just the "plasma" panel) and don't have any context menu or such for configuring them. I guess I need to take some deeper look into manual config.

With that, I'm basically back on a usable system as long as I have syslog turned off and don't need desktop icons. Here's a useful piece of eye-candy I have enabled now: the window list overview I have set as a hot corner action (when moving the mouse out the corner) on the upper left corner works nicely and is quite useful:



All those images are live views and update with whatever's going on in those windows, clicking on the window in that view goes directly to it in the respective virtual desktop.

GNOME apps like GIMP or XChat, which I use a lot, don't look too well yet, and the latter can't do it's transparent background stuff yet which I like so much, all graphics are slower than before (I already noticed the same with my laptop, so it's similar on nVidia and Intel video drivers, probably something with the XComposite stuff) and there's a few other small things that aren't yet prefect, but it looks like I have a pretty decent way of "Ling On The Edge" here again.

Time to get back to work.

(Tomorrow, at least. Tonight it's time for my favorite bar and some Karaoke. )

I just read an interesting post by Aaron Seigo from the KDE team.

As you might know, that development team made the choice to release their new technology, which was built to be up to the challenges of the future, once the backends themselves were stable and a good base for building applications on it, about a year ago, with the KDE 4.0 release.

A lot of bad press followed, as the applications themselves weren't completely ready for the users and the 4.0 release didn't even nearly ship feature-parity with the 3.5 series that provided and still provides a complete and stable user experience for many people. Now the next public beating goes round as Linus Torvalds has apparently switched to GNOME in the absence of his distribution being able to provide him with a KDE version that is feature-complete and stable for production use. They shipped KDE 4 as its said to be the new stable version and didn't offer people to stay with the actually production-ready KDE 3.5 (unlike openSUSE, by the way), but thankfully offer GNOME, which is production-ready, even though it doesn't offer many of the more future-oriented eye-candy and development frameworks. Linus apparently determined he'd better use a production-ready system than cool technology that is incomplete, as many other users might have.

This somehow reminds me of Netscape 6, we saw very similar patterns there. Cool new technology, but unfit for daily use of many people. Netscape died, and Mozilla needed a shift in thinking to really gain strength again, and the suite will never be a mass product any more, even though we finally managed to make it survive and hopefully thrive again with SeaMonkey.

Now, if they wouldn't have released the new technology in KDE 4.0, people probably wouldn't have put so much development into getting the applications and the user experience up to speed and be able to ship a really mostly feature-complete and for sure production-ready KDE 4.2 this upcoming week (I've tested and used it on my laptop and will convert my main machine soon). In terms of getting more people to work with the new technology, the "release early" strategy helped quite a lot.

The real question will be if the project and the KDE 4 technology can egalize the bad reputation it has right now, gain strength and move forward again as a strong platform and come back as the leading free desktop system. KDE 4.2 surely has the power to do that, but bad reputation is hard to overcome. Let's see how this turns out.

(And yes, when we have a certain reluctance to releasing SeaMonkey 2 before it's really production-ready, we also have stories like this in mind.)