[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Voting template

On Tue, Dec 12, 2000 at 10:18:23AM -0600, Brian McGroarty wrote:

> OpenBSD: A good, secure base. Beneficial in establishing a Debian
> distribution as the ultimate secure OS. Ported to a wide variety of
> platforms.

But it lacks many features found in FreeBSD, for example, such as SMP
and libc's locale support.
OpenBSD's strength is in secure userland tools (ftpd, openssh and all the
others). The kernel itself isn't secure or not -- it's just the kernel,
and have not that much improvments over NetBSD.
We can easily use their software to create a secure Debian system, but
here we must decide what kernel to use.

> NetBSD: Ported to even more platforms, more Linux-like
> development. Where OpenBSD has a focus on being secure above all else,
> NetBSD strives for a wide variety of features on a wide variety of
> platforms.

Yeah, this is the problem. The fastest, feature-rich and most used kernel is FreeBSD,
but it lacks multi-platform support.
In my mind we should build one usable system on x86, and then can proceed
to create a NetBSD-based multi-platform distribution.
After all, x86 is the most used platform around ;-)

> 	1. FreeBSD has commercial backing. BSDI stands behind FreeBSD,
> and it's not at all unreasonable to expect that they would embrace and
> support Debian/BSD.

That's one of the reasons why I think we should choose FreeBSD.

> The secure base speaks for itself. As to the developers: if Debian/BSD
> offers a superior ports system, it's entirely reasonable to assume
> that many/most OpenBSD developers will be quick to adopt it.

First of all, I think we shouldn't speak about any ports in Debian/BSD.
If we're building Debian, we should have packages.
In case we gonna use OpenBSD's software, we can work together with OpenBSD's
developers, and they would automatically help us.
As for the ports, they don't actually audit them -- only pre-built packages are revised,
and not that deep. They do most of their audit work on the
software written by themselves.

> Whichever base is chosen, the libc included is going to be well-tuned
> for that OS, and that OS for the BSD libc. The top half of the kernel
> and libc are pretty well blended.


> BSD would be more widely accepted by existing BSD developers. (Can
> anyone comment on Mac OS/X structure?)

I think we are not aiming to the BSD developers -- they are happy with
their BSD's. The main target is current Debian GNU/Linux users, and
all of them are already used to the current way of configuring the
As for the BSD world, they should know they're using Debian, and
they can be sure that if they move to another Debian, they'll 
be using the same configuring way.
The keywords of Debian (in my mind) is "standarts" and "unification",
and it is of no sense to break it.

> Native will perform better and will be more readily accepted and
> therefore supported by developers affiliated with the chosen BSD base.
> There's no reason to preclude Linux binary /support/, but it would be
> nice if these were flagged as such and could be relegated to their own
> tree, as non-free is now.

Agreed also. I don't see any point to even discuss this question, but
there has been many rumours saying that this is the way to run
Debian packages.

Regards, Wartan.
Sorry my English -- I am not a native speaker.

Reply to: