On 11/7/19 3:19 PM, Michael Richardson wrote:
Keith Moore <moore@xxxxxxxxxxxxxxxxxxxx> wrote:
> This is where the biggest disconnect between 2026 and reality is. If the
> reality is that industry is going to deploy implementations at Proposed
> Standard or sooner (and as far as I can tell, that's been reality for as long
> as there's been an Internet "industry"), it makes sense for IETF to recognize
> that and react accordingly.
You are saying this as if it's a bug.
It's not! It's by design.
We deploy at PS in order to find out if there is interoperability.
My first impression is that this is indeed a bug, a tremendous
disservice to Internet users. But I remind myself that automatic
software update is becoming increasingly common. So at least for
products that are certain to be run on Internet-connected hosts (as
opposed to, say, on air-gapped networks), and for which secure update
can be provided, there might be cases for which premature deployment
makes sense - especially if there's a strategy designed into the
protocol for dealing with version skew in the endpoints'
implementations. (Assuming public internet connectivity isn't at all
the general case, but it's an important corner case.)
Also, PS criteria were specifically not designed for deployment; the
idea was that you need interop testing before deploying. But if
interop testing were incorporated earlier in the process (which does
seem to be more common these days) then the sequencing assumed by 2026
might be suboptimal.
I could imagine that rather than the initial RFC being at PS, it could
be (for some cases, probably not all) at something akin to what Draft
Standard used to be - interop testing already done by the time the
initial RFC is published, with any changes made as a result of the
testing incorporated into that RFC.
> If we want there to be a prototype "just for testing" status, it should
> probably be called something other than Proposed - the name has come to mean
> something else in IETF context. And we should deliberately change one or
> more protocol elements to make the standard incompatible with
We do "just for testing" regularly at the internet-draft stage.
yeah, and it doesn't make sense to go through the whole RFC publication
process just to agree on a specification to test to.
Keith
