Date: Thu, 14 May 1998 02:18:34 +0200 From: Pierre Beyssac <pb@fasterix.freenix.org> To: "Jordan K. Hubbard" <jkh@time.cdrom.com>, Ollivier Robert <roberto@keltia.freenix.fr> Cc: core@FreeBSD.ORG, net@FreeBSD.ORG Subject: Re: INRIA IPv6 on FreeBSD Message-ID: <19980514021834.C6349@fasterix.frmug.fr.net> In-Reply-To: <6876.895046243@time.cdrom.com>; from Jordan K. Hubbard on Wed, May 13, 1998 at 12:57:23AM -0700 References: <19980513005824.B17879@keltia.freenix.fr> <6876.895046243@time.cdrom.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, May 13, 1998 at 12:57:23AM -0700, Jordan K. Hubbard wrote:
> We've gotten to the stage where we're at now through the various IPv6
> folks failing to talk to one another or look at the competing
> offerings, choosing instead to go focus exclusively on one specific
> favored implementation. That has to change or FreeBSD likely as not
Disclaimer: I'm not from INRIA. I ported their implementation to
-current in my SPARE time. It went easier than I expected so I
distributed the patches.
I'm "biased by design" since, from the start, I decided the IPv6
I wanted was the INRIA stuff. One of the reasons for this was that
it had been available for a long time and it was available both
for NetBSD and FreeBSD. Another reason is that I tried it more than
a year ago under 2.2.something and it worked pretty well for me.
So when I began to feel like I wanted to play with IPv6 at home,
I pretty naturally came to the INRIA stuff.
However I've had a quick look at the WIDE implementation from a
few weeks ago (when I began the INRIA porting). Back then, I was
under the impression that it was slightly less integrated with
common FreeBSD tools than the INRIA stuff.
The INRIA approach is to patch existing tools to make them
IPv6-compliant. This holds for sendmail, ftpd, ftp, telnet, rlogin,
inetd, for example. For tools for which this doesn't make as much
sense (ping, traceroute...), separate versions are provided
(ping6...). The INRIA code is available for NetBSD and FreeBSD.
The WIDE approach, from what I understand, is to provide new tools
for a lot (not all: some are integrated too) of these. They have
a separate inet6d for IPv6 daemons, for example. I think (but I
can be wrong) it only exists for FreeBSD.
The WIDE approach is probably better in the short term (less
integration headaches). The INRIA stuff, IMHO, is the way to go in
the long run (the ease of use of a common inetd is good illustration
of this). The INRIA stuff is apparently (as I said above, I'm not
aware of any ports of the WIDE stuff but it can be just ignorance)
available on more systems and more architectures (NetBSD) so they
have an advantage on portability.
> won't be adopting _anyone_'s IPv6 implementation because we're never
> going to get past the vague handwaving stage. Is it too much to ask
> that someone at INRIA take a look at the WIDE stuff in *detail* and
> vice-versa?
>
> Sorry to sound peeved, but it's been a frustrating exercise trying to
> get the various IPv6 camps to even consider working together and it
> seems that it's always been a "use OUR stack!" issue rather than
Jordan, I really think this is too much asking. I think it's
perfectly understandable that people working on an implementation
don't bother with details of the other implementation. They don't
have time to. Besides, comparisons from anyone working on a given
implementation would logically stand the risk of being considered
biased. The mere fact that you consider previous discussions on
the matter as "use OUR stack!" is very telling :-)
A while ago, someone on freebsd-security (quite sorry I forgot his
name) said he had written (or was working on) a report on performance
and implementation issues after comparing both implementations. I
don't know what it became. IIRC, his conclusion was that the INRIA
stack was better. Not having seen the report, I'm sorry I can't
exactly say how (on what measurements) he determined that.
> "we'll get together,work something out and get back to you" with IPv6.
> This wouldn't fly in any other area of FreeBSD development and I fail
> to see why IPv6 should be any different. You folks have GOT to start
I don't think the ball is in INRIA's or WIDE's camp. They have
working implementations, they maintain these according to the
evolution of IPv6, that's their goal, period. We can be glad and
proud they considered FreeBSD a good platform for their experimentations,
but I don't think we can ask for more.
Obviously, having a single implementation to choose from would make
FreeBSD's choice easier, but that's not gonna happen unless WE
(FreeBSD) decide to work on it for our own purposes. _BUT_ I honestly
think that reinventing yet a new wheel is a complete waste of time
except for people pursuing religious concerns such as "BSD network
code sucks so I write my own, cleaner code so I can mess up with
socket semantics and ICMP includes" (this is a purely theoretical
example, obviously :-)
In conclusion, anyway, the ball is in OUR camp and unless we realize
that, we're gonna be late for the IPv6 boat.
--
Pierre Beyssac pb@fasterix.frmug.org pb@fasterix.freenix.org
{Free,Net,Open}BSD, Linux : il y a moins bien, mais c'est plus cher
Free domains: http://www.eu.org/ or mail dns-manager@EU.org
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-net" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19980514021834.C6349>