Subject: Re: The Good, The Bad, & The Ugly
To: None <current-users@NetBSD.ORG>
From: David Jones <email@example.com>
Date: 05/25/1997 09:32:41
Jukka Marin wrote:
| On Sun, May 25, 1997 at 05:23:51AM -0500, Tom T. Thai wrote:
| > I found a site that did a comparison of NetBSD and FreeBSD. The link is
| > at: http://www.cons.org/cracauer/bsd-net-vs-free.html
| Didn't sound to good. Wish he could at least spell 'NetBSD' properly. ;-)
I don't see what's so bad about it. Martin's pretty truthful here.
Our VM system DOES suck, and there's a PR on file whose last word is
"rewrite VM system". Obviously not something you can do in a weekend. OTOH,
bug fixes for the worst 1.2 VM problems were committed, so this isn't so
much of a problem.
As for using vi over NFS, I've done that and haven't lost. I think the real
losers here are the programmers who assume that VM and buffer cache are
coherent (possibly because they're merged). That just ain't so over NFS
on _any_ system (even SunOS). Been there, done that. mmap() isn't as hot
as people think it is.
A lot of the source changes are cosmetic, and don't add functionality.
Perhaps this will result in a better system in the long run, but it does make
it a pain in the ass to compile things. You really have to sup and compile
the whole thing and "keep current". I've often tried to upgrade only parts
of the system (those whose bugs I've come across) with mixed results.
If I was running 1.2-release and had to upgrade a C file because of a bug
I reported, fixed in -current, then would I have to upgrade to -current?
It also makes it hard to maintain packages. How many times have we imported
IPfilter, and then made our NetBSD-specific cosmetic changes?
If I had to start all over, and I had only PC hardware, I'd probably install
FreeBSD too, since it is more user-friendly, and the kernel code is closer to
the original 4.4BSD sources. But I have an Amiga and a Sparc, so it's either
NetBSD or Linux. Easy choice. :-)