Re: [dtn] TCPCLv4 comments

["R. Atkinson" <[e-mail]>]

>  - I am absolutely fine with making TLS mandatory (my own use of TCPCL is to always require TLS),
> as it would simplify logic. Some may dislike this, but if it's easier for approval then I would prefer that.
> There would be no prohibition from somebody implementing a "TCPCLv4 without TLS" as a
> separate thing for closed networks.

The text quoted above is ambiguous, so I can’t figure out what it means.  


The IETF has for about 30 years now taken 2 complementary positions on security:
  A.  Security has to be specified and Security has to be “MUST implement”, so that is is available to use in all shipping products.
  B.  Security is OPTIONAL to deploy/use and this is left as a decision for the network operator or user.


The IETF generally avoids telling operators what they have to do.  Yes, we do have a growing
number of “Best Current Practice (BCP)” documents, but that is in the category of strong advice,
rather than a standards mandate.

I concur with the long-standing IETF position on security.  

So in the context of TCPCL, I believe it is very important for TLS to be “MUST implement”, 
but I object to any notion that EITHER “standards-compliant deployments” OR “interoperable
deployments” MUST deploy/user” TLS.

In short, it needs to be possible to interoperate fully - whether or not - one happens 
to *use* (“deploy”) TLS  for a particular TCPCL session.  

Yours,

Ran