Ex-Microsoftie: Free Software Will Kill Redmond (ComputerWorld)

"What is bad is going to become better"

Seems unjustified coueism to me.

Maybe in the bells and whistles world of Gnome, KDE and the others this is the case -- and its probably mostly true for Linux as well, but my personal feeling, being a Debian user for 10 years, is that too often very basic tools and legacy software that used to just work tend to break because of inconsiderate changes in the environment -- for example try to use OOo under Window Maker : used to work 3 versions of Debian ago. Not anymore. Even "new" fashion tools are sometimes completely broken (beyond repair?) (thinking of NetworkManager for example -- or I once took a look at almost every famous graphical music players and managed to make half of them crash in less than 3 minutes by just using my mouse, while the other half needed a PhD to understand their interfaces... we used to have xmms several years ago -- now there is nothing usable and stable resembling to that and xmms is not packaged anymore)

Another example : right now I'm experiencing delays that look like random in the order of 0 to 1000ms when using "less" or "mc" through ssh on a Etch server using a Lenny client (with rxvt terminals under Window Maker). Never saw anything like that before, and do not even know where to start investigating to fix this stupid behaviour. I'm not yet annoyed to the point I really care about it (I have a lot of other things to do -- including long rants to write on LWN :P ) but it is still a very frustrating experience, especially because I never saw a such utterly broken behaviour before. So it feels like (and I don't know if this really is the case or not) somebody has cluelessly broken something in ssh or in less or in rxvt or in X window or in the X protocol or somewhere else in the string of components that should be robust I'm using.

All of this is a pretty dark picture -- passing under silence brighter points of GNU/Linux and associated graphical environments -- but working all the time under such a system I begin to be a little tired of people that just think that it is either good or will be. It is sometimes not. It is sometimes even very bad -- and it sometimes even become worse. So in my experience the bad is not always becoming better -- and I'm far from being sure that the "process" that lots of packages adopt and even some popular distributions can help ; too often the "process" is simplified to : mix bugfix and new buggy features, then ship the whole thing to the world for testing, then start over -- I don't see the point where stable software can emerge from such a process, indeed even if you remove the "add new buggy features" part you will still fail to produce something of decent quality for big packages without at least proper regression testing, which is seen far too rarely.