Re: [tcpm] Secdir last call review of draft-ietf-tcpm-rfc793bis-24

[Kyle Rose <[e-mail]>]

On Sun, Sep 5, 2021 at 12:51 PM Gorry Fairhurst <[e-mail]>
wrote:

> To me, this has a much more complicated history, and I think TCP has been
> extended many times - albeit not sucessfully in the ways mentioned above,
> but in other watys. It seems like a discussion of whether ossification has
> been good or bad. I'd also really quite concerned to see words like "often"
> used without clarifying further -  there are legitimate cases where
> filtering can be useful for managing the security of TCP connections: A
> firewall in one context might do many things, and that might actually be a
> good security model; in another context that might be different.
>
I'm not in favor of making a value judgment or discussing the wisdom of
approaches to addressing ossification in this document, but I do think,
given the rare opportunity of publishing a 793bis, that it's worth putting
the implications of TCP ossification in writing so we don't have to
relitigate these issues or explain in detail why such-and-such won't work
whenever they come up in future attempts to extend TCP. I guess I'm arguing
that we shouldn't need Joe Touch to explain multiple times to multiple
groups why a given proposal won't work when we can explain it once and
mic-drop a one-line reference to that.

I'd also question the importance of
> https://datatracker.ietf.org/doc/html/draft-iab-use-it-or-lose-it, in
> relation to the core TCP spec, even though the message is clear for MPTCP.
>

This is why I put it in a parenthetical. It's not directly relevant to TCP,
but it does indicate to the interested reader an approach that newer
protocol designs have taken to avoid one of the causes of ossification. I
agree it's not essential.

Kyle