The Importance of Being Linux (ABC)

Twice a year there are new releases from the major distributions. The included software is usually vastly improved over the previous ones, even if the distribution itself hasn't changed significantly. If you chart back and examine the development of OSX or Windows, it feels like Linux Distributions have come along further and faster than any of these in any period of time. And the pace seems to be getting even faster now. We are seeing regularly scheduled releases from projects like Gnome, KDE, Mozilla, OpenOffice, and many others I am always using.

Of course, you could point out several stinkers as far as development speed. Things like the Hurd kernel are probably the biggest offenders. I can't decide to laugh or weep at the Debian release cycles.

But overall I'd say Open Source delivers more in less amount of time.

As for using fear as a way to get people working.....

Wasn't the whole Tolkien trilogy about getting rid of people like that, and letting people work and live based on their love of others?

The idea of a world where fear motivates people to accomplish somthing is about the sickest idea I have ever encountered, and, frankly, people who use that technique are the ones who make me wish I didn't have to go to work in the morning at all (and that I could do something to make them go away, and then enjoy my work again.)

In fact, I will say that my most productive years were those in which I felt very *little* fear, and was greatly appreciated by my bosses, who were not mean.

If anyone asked me for advice about dealing with a boss who motivates by fear, my first suggestion would be to leave the job and find another one; my second would be to try and find a kinder person over that boss, and my best suggestion would be to somehow get that boss in trouble with the police if possible (because someone who uses fear as a motivator is probably doing something else that's very, very wrong.)

When I used it two years ago, JBoss was, as far as I could tell, perfect. Has it changed? Not so far as I can tell. Has Apache changed in the 8 years I've been using it? No, it worked right then, and it's the same now.

Of course, there are plenty of projects which show great improvement over a short period. But the projects that people think of are generally the ones which are already good, and there's generally not much to do to them.

The kernel changes because there were (and are) things wrong with it that aren't a big problem for the current users, but are a problem for a lot of potential users. 2.4 was no good if you needed a list of all of your known devices; the current code has problems with more than 256 processors; there are always new devices which need support.

Desktop software changes because it's an area that isn't well understood. Gnome and KDE (and Windows) are popular not due to being right, but due to the right thing having not been worked out yet.

The reason the article comes out positive, which I'm not sure the author realizes, is that he uses a lot of software which is perfect for his needs already. When he says that he wants Banyan Vines to be open source, I bet he doesn't have any changes to it in mind, aside from having it ported and built for his current computer.

Being a technology columnist seems like a good racket, apparently you can write anything you want and they will publish it. I should look into it when I get tired of working for a living.

ESR once coined the mantra "Release early, release often". (Sorry, too lazy to look up the full citation.) I would rephrase that as "Give people access to your development sources." There are many very good open source systems that haven't released a new stable version since years. Sometimes this is a Good Thing(tm), sometimes some interim version would be nice. Debian comes to mind, of course; but also less prominent examples like ht://Dig. Such stable versions are often released more seldom than their proprietary counterparts. I'm active in LaTeX development, so I know about late releases quite well... :-) Of course, this does not really matter for Open Source pundits; they can access the development version. For them, the system changes often.

On the other hand, for companies it's sometimes hard. Many open source policies in companies state that one may only use stable releases. From the view of such companies, there is also an other issue: They often get their software package updates from vendors, together with support. So the time gap from development to their vendor is often even more important for them. And we need only look at the security section of this LWN issue to learn about some update delays at distributors. ;-)

So, IMO open source has both properties: Fast update cycles, and slow release cycles. And that's OK, for most of the time. In companies, we don't want ever-changing software. We want, need, and get some stability. Sometimes stable releases might come a bit faster, but what the hell. There's the larger problem that it's sometimes hard to get support for older releases (say, 3 to 4 years old) that are still in production.

Disclaimer: I do consultancy work for executive management of multi-national companies. Part of that work is also on open source strategy and open source policies. That opinion above has been built by analysis of major open source systems I have recommended or wanted to recommend to my clients, and sometimes couldn't.

Cheers, Joachim

Also, i _use mh, and i want to _keep using mh, so go have a beer, pat yourself on the back and then get busy!

"Release early, drink 'till closing"