Re: [clue] [art] ICE, ICE-bis, and Cluster 238

[Adam Roach <[e-mail]>]

On 9/6/18 3:24 PM, Cullen Jennings wrote:
> I would propose that instead of the parts of specs that use the 
> trickle mechanism purely reference that part of the new ICE spec that 
> defines the syntax and say to transfer trickle candidates.


Unfortunately, it's not quite so simple. The mismatch with trickle-ice 
is only the most obvious problem. In practice, there are a few dozen 
places where unpublished documents in the cluster refer directly to one 
version of ICE, and indirectly to another, or indirectly to both.

To help wrap our collective minds around this issue, I've put together a 
dependency graph that shows which Cluster 238 documents refer to which 
version of ICE. The red lines indicate references (directly or 
indirectly) to RFC 5245, while the purple lines indicate references to 
RFC 8445. The thickness of the line indicates how many hops away from 
ICE the reference is. Dotted lines are informative references.

https://trac.ietf.org/trac/art/wiki/c238

To put a finer point on this, it took me a couple of hours to pare this 
graph down to something that was small enough to look at and actually 
understand -- there are actually twice as many documents involved in 
this mess, but including them makes the situation too confusing to even 
start to get a grip on.

Everywhere you see a purple document with a red arrow or vice-versa is a 
place we need to figure out how to handle this mismatch. You can push 
the boundaries around, but the count of mismatches is pretty much the 
same no matter how you do that. On a first order estimation, this is 
days -- or, more likely, weeks of work. And I'm talking about 
person-hours, not calendar time. If you can do this, or find someone to 
step up and do it, then your proposal would be a viable alternative to 
the one that the ART ADs have proposed. Short of that, I'm not sure how 
we can do what you want.

/a