BSD development is conservative. Before new additions to the base system are available as a part of STABLE (production release), they undergo severe testing; therefore, BSDs lack a great variety of flaky drivers and questionable stuff that is all arond the Linux kernelYeah. Remind me of one "flaky driver" or "questionable stuff" that was recently added to a stable Linux kernel release. The 2.4 maintainer is EXTREMELY hesitant to include anything new, to the point of frustrating people (see XFS).
BSDs have a broader sense of the base system. In particular, BSD integrate kernel, libraries and some binaries together to make the base.Give me a break. Are you familiar with GNU/Linux? Are you telling me that the development of the GNU C library and POSIX userland are not closely following the development of the Linux kernel? If it weren't for Linux, what else would they bother with? The Hurd is decades away.
You could say that the idea of a "base" where the same group of developers work on all the code provides greater coherence to the user, because you don't have different developers pulling in different directions with respect to the usability of the base system. But that is an aesthetic preference which you are pretending is a practical one.
Closer integration means more polishing; that leads to greater stability.Please indicate what you mean by "polishing", and why it implies greater stability. My opinion is that loosely coupled code can be perfectly stable as well, if one follows proper software engineering principles in its construction.
The bonus that you get with loose coupling is a greater degree of reusability (hence, the GNU userland runs on just about every Unix-like kernel on the planet). But that in no way implies that the development of the GNU userland with respect to Linux is a second-rate effort. The fact is that Linux is where the money and the fame is today, so Linux gets first class treatment, but the aims of the GNU project ensure that other platforms are not left out in the cold.
I don't understand why the FreeBSD ideal (one kernel, one libc, one
If you go through the configuration file and comment out everything that you do not need, you will have a very tiny kernel. That can increase a chance of having longer uptimes.Riiiight. And pray tell, if one built a bunch of modules but left them unused, how does this affect stability one bit compared to a kernel where the unused modules were pruned out by hand in the configuration? Sorry, I don't buy it. System stability is unrelated to code that is available but never called into.
When Linux developers try to include an absolute enormous amount of hardware support provided by default kernels, BSD developers provide only what is needed for basic functionality; that is truly a big plus.Have it your way. I think being able to use my hardware (even with a beta-quality driver, if forewarned) and send any bugs I find upstream is "a big plus". But then again I like having useful computers, as opposed to engaging in minimalism zealotry as some seem to prefer.