On 07/04/2013 01:11 PM, Francis Dupont wrote: > In your previous mail you wrote: > >> Considering the above, I guess I'm in the camp of "avoid fragmentation >> where possible". However, I don't think I'd go as far as deprecating it. > > => I have private notes about the deprecation of reassembly so > not so far... BTW there was (still is) a TCP/IP stack so poor it > doesn't handle reassembly of out of order TCP fragments, so the idea > is not so silly.
If you're going to deprecate something on the assumption that some implementation does something stupid about it, then I wonder what we'd be left with. -- for instance, you should be deprecating Neighbor Discovery, because many implementations do things such as allowing the NCE to consume all memory, not doing any sanity checks on the received link-layer addresses, etc., etc. Cheers, -- Fernando Gont SI6 Networks e-mail: [email protected] PGP Fingerprint: 6666 31C6 D484 63B2 8FB1 E3C4 AE25 0D55 1D4E 7492 -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
