But let's just take the migration away from CVS as an example. The current view on the topic is that no existing VCS can replace CVS and keep some of the properties that CVS has. Therefore, if we were to switch away from CVS, a bunch of developers attached to those old properties would possibly abandon the project.
A major open source project fighting about keeping CVS? Wow.
FreeBSD still distributed their ports tree over cvsup up until about a year ago, even though the primary workflow was moved off of CVS quite some time ago.
I guess he's writing for die-hard NetBSD users, because he doesn't make much of an effort to explain why anyone should care about the project anymore. (I would be interested to know why.)
This rings eerily similar to the state of the Gnome foundation and other long-in-the-tooth FLOSS projects. Sad to read but I really don't know how one responds to "no innovation." Is "come up with good ideas" really a thing?
I can't think of a good reason to use NetBSD [1] over Linux [2] for any architectures that Linux supports. I guess licensing is an issue for some companies. If there's another reason, the NetBSD project needs to communicate it better to attract new users.
You can use both - you can run NetBSD in userspace under Linux http://www.netbsd.org/docs/rump/ - eg to run userspace filesystem drivers or a userspace IP stack... you can't do that with Linux.
I have very recently started using NetBSD, after having been interested for a while (had a tendency to go to the Fosdem talks on it for a few years). I think defining what the aim is (other than "runs anywhere") would be helpful; as would not cvs. All the BSDs have a user issue, with the dominance of Linux, and they have to find larger niches.
Yes, but so does Linux. Actually, given the tremendous industry support for Linux and the increasing obscurity of NetBSD, it wouldn't surprise me at all if Linux ran on more architectures now...
In terms of other hardware I believe it bends even more in Linux's favour. I'm not sure if any Linux distribution has as many ports as NetBSD though. Gentoo and Debian probably come close.
What you've said doesn't even agree with the NetBSD article you linked to (which suggests 15 processor architectures).
Even if you could create figures that seem accurate, you really lose some of the nuance with a comparison like this. Questions like which OS you're better off with if you're attacking an architecture without an MMU (linux for sure) or which software distribution treats your old SPARC workstation, for example, as a first-class platform (OpenBSD or NetBSD) instead of an oddball orphan seem more interesting.
Similarly, it'd be interesting to know WHY so many new platforms are using Linux instead of *BSD for embedded stuff. Is it something inherent to Linux, just momentum, driver vendor lockin from the vendors, name recognition, or what?
A major open source project fighting about keeping CVS? Wow.