"The next-generation of the KDE open source project, version 4.0, has been touted as the beginning of a new era in desktop computing, but only two months from the first release some users are wondering if it's just all hype. KDE 4.0 will bring a collection of new technologies to the Linux and Unix desktop, but there are uncertainties around how much of it will be included in the initial release. KDE user Andreas Pakulat expressed doubt about how the release will take shape in a blog post titled 'Where's the KDE4 desktop?'." KDE's Aaron Seigo also shares his thoughts on this.
To view parent comment, click here.
To read all comments associated with this story, please click here.
Member since:2005-07-07
Indeed. In the IT business, you have to stay fresh. Companies understand (understood) that too - Microsoft used to almost rebuild the company every 18 months. Everyone lost his position, it was all re-done. It kept them competitive. (now, of course, they don't need to be competitive anymore, they just need lawyers to keep their monopoly).
Member since:2007-06-06
Do you mean in KDE? If so, there is *massive* code re-use - it's one of the hallmarks of the KDE project.
Not KDE specifically, just the theme in large projects in recent years.
KDE4 isn't a re-write of KDE3, though - it's mainly a port to Qt4, along with legacy stuff being retired and approaches that turned out to be sub-optimal removed or fixed, plus the addition of a few new frameworks made from whole cloth
I never said it was. What i'm saying is the general approach of massive changes after years of development is flawed in most cases.
Plus, the difference in release dates of KDE3.0 and KDE4.0 is approx 5.5 years - rather a lot more than "every couple of years!" 
QT4 wasn't even available 5.5 years ago the first release was on June 28, 2005 - 2 years ago, which by definition _IS_ "a couple". KDE3 was both the stable and development release before then.
[/i]Giving them a chance at a clean break like this is tremendously liberating - they can tear down all of the cruft and poor design choices they inevitably accumulate and take their time doing it - it feels like you're being given a fresh new start with a clean slate, and you can finally go nuts! [/i]
The point it's far more effective to make an extensible framework, gradually improve it where there are limitations. Rather than powering through and creating yet more poor design choices in haste. This is much less to do with KDE4(which as you said is a port to QT4) and much more to do with QT4 but the theme remains the same in many projects. Taking a clean break every 2-3 years is not always the good thing you paint it to be. Having a base that doesn't NEED to make a clean break is far better.
Member since:2007-01-04
Member since:
2007-08-03
"There is very little code-reuse."
KDE4 itself is expected to last at least another 5 years.
Do you mean in KDE? If so, there is *massive* code re-use - it's one of the hallmarks of the KDE project.
"I'm not saying KDE4 didn't need to make the changes it did, because it likely did(especially with QT4). What i'm saying is the whole approach of rewriting everything every couple of years is massively flawed, slower and bug ridden."
KDE4 isn't a re-write of KDE3, though - it's mainly a port to Qt4, along with legacy stuff being retired and approaches that turned out to be sub-optimal removed or fixed, plus the addition of a few new frameworks made from whole cloth, most of which had no direct suitable analogue in KDE3. Bear in mind that KDE is largely a volunteer project and that its SVN contains somewhere in the region of *5 million* lines of code - a re-write would be a practical impossibility!
Plus, the difference in release dates of KDE3.0 and KDE4.0 is approx 5.5 years - rather a lot more than "every couple of years!"
I think having a large break with old code is massively beneficial psychologically, too - programmers seem to get more and more worn down when they get stuck working on the same code that is expected to continue functioning in the next point release - there's a tendency to become more and more conservative and resistant to change in case something breaks and people whinge, and development becomes a chore. Giving them a chance at a clean break like this is tremendously liberating - they can tear down all of the cruft and poor design choices they inevitably accumulate and take their time doing it - it feels like you're being given a fresh new start with a clean slate, and you can finally go nuts! The introduction of better and more powerful APIs (for which retiring old and, with hindsight, ill-conceived ones is often a pre-requisite) makes the process of going nuts even easier! In a nutshell, I think these massive shake-ups are very healthy and desirable indeed as long as they are paced correctly, and having one every 5 years feels like a pretty good tempo.