Reasons for the removal include that I don't use NetBSD in production anymore and that I am now much more experienced in FreeBSD so that I find solutions faster in FreeBSD, no matter how good the relative quality in the area in question is.
Most problems in my original review have been fixed anyway, although I am much happier with how FreeBSD turned out (the VM system is still much faster than NetBSD's UVM, although the latter at least solves the bugs and I weigth SMP over multiplatform). Portability of other OpenSource software is not an issue anymore, that FreeBSD has the edge in availability of binary-only software or highly specialized packages (especially programming languages like CMUCL) should be obvious.
Overall, I think both are excellent and - even more important - actually attack the problems that bother me. Whereas Linux in comparison develops fast as well, but on one hand doesn't address some of its fundamental issues (kernel memory management, middle SCSI layer, booting, still asynchronous metadata in default filesystem - and lying about the effect of it, switching to non-overcommit memory as a default) and on the other hand weigths the value of crap drivers for crap handware heigher than the cleanup that could be done when overly intrusive drivers were removed. In the BSD world, drivers behave well or they are out at the next release. Also, it becomes more and more visible that the GPL (GNU license) is a serious problem. I also have a Web page about this at http://www.cons.org/cracauer/gpl.html. I won't even start of chaotic driver update mechanisms (ISDN, PCI and a few more), missing version control, missing change log messages and everything else that most other Opensource projects have these days.
I felt that listing issues in NetBSD and FreeBSD that are minor in comparison doesn't make senses and leaves readers with a wrong impression.
$Id: bsd-net-vs-free.html4m,v 1.2 2000/09/11 09:45:14 cracauer Exp $