Re: [tcpm] Flaw in RFC793 (Fwd: New Version Notification for draft-gont-tcpm-tcp-seq-validation-03.txt)

[Fernando Gont <[e-mail]>]

Hello, Lars,

On 03/06/2018 04:05 PM, Eggert, Lars wrote:
> Hi,
> 
> isn't this something that could simply go into 793bis, after there is consensus that something needs to be fixed?

(me thinking out loud) Dunno. My understanding was the rfc793bis was
only taking documents that had already formally updated rfc793.

A "feature" of that is that the rationale for the changes incoporated in
rfc793bis are documented in the updating documents, and rfc793bis
doesn't need to delve too much into such rationales, or explain the
changes -- just incorporate them.

What's in draft-gont-tcpm-tcp-seq-validation is certainly necessary to
illustrate where RFC793 fails. But I doubt you'd like the time-sequence
diagrams of all these corner-cases in RFC793bis. But you'd also like to
have something that you can point to explain these changes (even more,
if different popular implementations decided to address this in a
different way).

One possible way would be to add all this to an Appendix in RFC793. But
I guess that would warrant to do the same for other stuff (e.g., update
for the TCP URG mechanism).

Thanks!

Cheers,
-- 
Fernando Gont
SI6 Networks
e-mail: [e-mail]
PGP Fingerprint: 6666 31C6 D484 63B2 8FB1 E3C4 AE25 0D55 1D4E 7492