Great read. I think the biggest takeaway is how they organize and run their operation. There's a tiny amount of high talent and dedication people that really control the process. They're open for contributions by whoever wants to try. They internally filter the wheat from the chaff to find those who contributions are truly useful. They give them extra power in both what ends up in the distro and helping the filtering/discovery process. The rest are still allowed to attempt whatever contributions they can. The model seems to hit an optimal point to bring in more good than bad.
Might be worth copying in other projects. Supporting this is the other, highly-successful organizations that use a similar model. Far as Linux, there's always exceptions to the rule and lets not forget it depends on GNU work done differently. That hybrid, immensely popular development doesn't compare apples to apples with about anything. I treat it like its own category.
Interesting read, thanks for posting. Brings a bit of relevant historical perspective given that I recently switched to OpenBSD for a project I'm working on.
It is funny how USL vs BSDi hurt BSD at first, but then SCO targeted Linux not BSD for a reason. Terry Lambert has a lot of posts on the backstory behind the settlement BTW (clue: you can thank Ray Noorda).
We know. The culture in Unix was always sharing source and user collaboration. When AT&T finally tried to exert control there was a large movement within the community to go free of AT&T code. GNU and BSD were both a reaction to that.
The rise of x86 and the availability of free Unix (in the form of BSD or GNU) was destined to destroy the proprietary Unix market. In 1991 neither BSD nor GNU had kernels booting on x86, leading Linus to eventually release Linux in August.
Linux had the luxury of being the only readily available Unix-like kernel on x86 for several years.
The USL v BSDi lawsuit slowed BSD efforts for nearly two years while Linux gained mindshare and features (specifically x86 features & drivers). Once the suit was settled out of court development took a long time to regain momentum.
Net/2 was released in June of 1991 (before Linux!), so if the USL lawsuit hadn't happened, we might all be running that instead.
That is just playing futurology. We will never know.
Just like some of us would have liked that Apple would be just as successful with BeOS instead of NeXT, but we will also never know.
In 1993, I learned UNIX on Xenix system. Coherent being the other option.
Linux only became production quality when the likes of Intel, IBM and others started supporting its development. As far as I am aware they never supported BSD like that.
The really nasty bugs are found by a couple
of really smart people who just kill themselves.
Most people looking at the code won’t see anything
How many people looked at heartbleed and didn't see anything? How many people looked at goto fail; goto fail; and didn't see anything? And those were relatively simple. I wonder how many "really nasty" security bugs are still out there?
Might be worth copying in other projects. Supporting this is the other, highly-successful organizations that use a similar model. Far as Linux, there's always exceptions to the rule and lets not forget it depends on GNU work done differently. That hybrid, immensely popular development doesn't compare apples to apples with about anything. I treat it like its own category.