Date: Wed, 24 Apr 1996 10:34:56 -0500 From: Jim Fleming <JimFleming@unety.net> To: Poul-Henning Kamp <phk@critter.tfs.com>, "'Garrett Wollman'" <wollman@lcs.mit.edu> Cc: "current@FreeBSD.ORG" <current@FreeBSD.ORG>, Jim Fleming <JimFleming@unety.net> Subject: RE: SLIP: Check IP Version (please test) Message-ID: <01BB31C9.AF275A80@webster.unety.net>
next in thread | raw e-mail | index | archive | help
On Wednesday, April 24, 1996 9:58 AM, Garrett Wollman[SMTP:wollman@lcs.mit.edu] wrote: @<<On Wed, 24 Apr 1996 10:09:07 +0000, Poul-Henning Kamp <phk@critter.tfs.com> said: @ @>> An IP v8 packet MUST never be forwarded to the slip driver. @> Fine, well, and where in our code do we assign IP protocol versions @> to interfaces ??? @ @We assign addresses (and their associated address families) to @interfaces, which amounts to the same thing. Any implementation of @IPvX where X != 4 MUST use family values other than [PA]F_INET. @ @-GAWollman @ This may be the general case. In the case of IPv8 it was intentionally designed to allow incremental "hacks" to be added to existing stacks without breaking everything in the world. Because of this, a new address family may not be needed. There are several benefits of this incremental approach. 1. It gives "students" the ability to try some simple experiments with their systems. This allows them to gain confidence that they can do kernel work (a lost art) without destroying the system the first time out. 2. It allows sites to use the IPv8 format for security functions in their closed systems that may need extra protection from IPv4 sniffing. 3. It allows developers to test the benefits of having the IPv4 core network still route some traffic while the IPv8 network is being developed around the edges using Protocol 4 IP-in-IP encapsulation. 4. Performance differences can also be tested when IPv8 packets are sent via the same transports as IPv4 with minor tweeking to allow both formats to co-exist. 5. Most importantly, it allows us to evaluate the trade-offs that were made in the current IPv4 implementation of putting upper level protocol decisions in lower levels of the stack for efficiency reasons while sacrificing code clarity. Now that systems are faster, some of these trade-offs may be relaxed in modern IPv8 implementations. ...there are many ways to skin a C+@...IPv4 did it one way...IPv8 builds on that... -- Jim Fleming UNETY Systems, Inc. Naperville, IL 60563 e-mail: JimFleming@unety.net
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?01BB31C9.AF275A80>