Hi Melinda,
On 10/31/17 2:07 AM, Melinda Shore wrote:
> On 10/30/17 2:03 PM, Russ Housley wrote:
>> There was not a document that describes the rsync protocol. The code
>> has lots of comments, and it supports many versions. The code is the
>> real protocol definition.
> I'd like to think that ultimately we'll be able to include
> code or other non-IETF-standard material in normative
> references, but I think there are some key issues not
> addressed in the draft, particularly around the stability
> of the reference. I think that revision control systems
> probably can be used to address some of that ("branch
> <whatever> on date <whenever>") but stuff gets moved in and
> out of repos all the time and it's difficult to guarantee
> the stability of material under someone else's control.
To me, this is where intent comes in. To put this in the negative,
because I think that's where we derive most value from the statement, if
someone doesn't intend an interface to be stable and mature, we should
not be referencing it. I think that accounts for over 99% of all
interfaces. Thus, starting with a statement that the interface is
intended to be stable is a big step. At that point what should we be
looking for and how should we judge?
Eliot
Attachment:
signature.asc
Description: OpenPGP digital signature