Re: How would I get debian unstable?
On Thu, Jun 05, 2008 at 06:13:51PM -0400, Douglas A. Tutty wrote:
> On Thu, Jun 05, 2008 at 10:44:01AM -0400, hendrik@topoi.pooq.com wrote:
>
> > I believe Berkeley BSD is free. I'm told Apple chose it over Linux as
> > a basis for OS/X because it didn't have all the restrictions of the
> > GPL that required it to remain free, though. I'm also told it is
> > developed less aggressively, does not support all the latest hardware,
> > and is more stable than Linux. Not that anyone can really call Linux
> > unstable.
>
> The BSD license allows distribution of binaries only for commercial use.
> Apparently if you install something called unix services for Windows and
> run strings on all the Windows binaries, you'll find a whole slew of BSD
> license copyright statements. This is a good thing since if TCP/IP
> wasn't licensed under BSD, there would be no internet; there'd be some
> microsoft network; there'd be an IBM net, an HP net, until someone did
> it under the BSD. Luckily, DARPA in effect hired USC-Berkley to write a
> network stack that could be used by different computers.
>
> I think the FreeBSD and OpenBSD people would argue with the "developed
> less aggressively" stance. OpenBSD folks do their development on new
> laptops. They release every 6 months but their -current (our Unstable)
> is never supposed to break and is perfectly fine in production; the only
> downside to -current is that there are no pre-built binary packages (use
> still use pkg_add but it ends up compiling the port instead of
> installing the package). NetBSD does seem to be developed at a slower
> pace. The big difference between FreeBSD and OpenBSD are that OpenBSD
> runs on more hardware and will not put binary-blobs or non-BSD code in
> the kernel whereas FreeBSD will do both. Hense, some drivers in FreeBSD
> are written by the hardware vendors (or others after non-disclosure
> agreements are signed) whereas OpenBSD (which often supports the same
> hardware as well or better) writes its own drivers via
> reverse-engineering the hardware if all else fails.
>
> As for stability: look at the debian packages it would take to make
> OpenBSD base install. At least a kernel, apache, shells, Xorg, standard
> tools, compilers, perl, lynx, ssh, ftp server, shorewall, various
> archivers, etc. Plus all their dependancies. Now look at the list of
> security updates to Debian Etch (yet alone Testing or Unstable) in the
> past six months. Now compare the number of security updates to OpenBSD
> in the same time-frame.
>
> Note that the reason that OpenBSD can claim only two security holes in
> the default install in the past 10 years is that there are no services
> active in a default install (you have to add commands to the startup
> script to enable them).
>
> People reoutinely built appliances like routers using OpenBSD and e.g. a
> Soekris box and put it on the shelf. They may only update it when a
> security bug happens (rarely). Since there are simple HOWTOs for making
> OpenBSD on a CF card, updating the appliance consists of swapping the CF
> card.
>
> I would call a box with a kernel security hole at least potentially
> unstable. Its been 10 years since OpenBSD had one, its been what, 2
> weeks, since Etch last had one? On this basis alone, I'd call OpenBSD
> more stable.
>
> > By the way, there may well be other systems that should be mentioned,
> > and I'd appreciate corrections if anything I've said is wrong.
>
>
> I hope my corrections and amplifications are generally correct.
Thank you. I wasn't aware of the way BSD has split up into different
streams.
-- hendrik
Reply to: