[rfc-i] For v3: Better identification for multi-document sets

Nico Williams nico at cryptonector.com
Sun Dec 29 13:08:15 PST 2013


On Sun, Dec 29, 2013 at 3:00 PM, Paul Hoffman <paul.hoffman at vpnc.org> wrote:
> On Dec 28, 2013, at 1:28 PM, Jim Schaad <ietf at augustcellars.com> wrote:
>
>> Another issue that needs to be considered is the question of recursion.
>> Should I be able to do
>>
>> <reference anchor="Keywords">
>>  <reference anchor="BCPXXXX">
>>    <reference anchor="RFC2119">
>
> -1. Although it allows grouping for multi-document sets, it also allows all sorts of weird things that we would never want. We would have to create fairly complex rules to avoid those.

Usages we don't want don't have to be permitted just because the
schema makes them feasible.  These can be forbidden at some other
layer.  Though I agree that it's better to formally express (e.g., via
schema) just the rules we want; I don't expect perfection.

> At this point, the requirements (not the design) are:
> - Ability to name references however the author wants
> - Ability to create a single reference that refers to multiple named documents
> - The tools that create the non-canonical formats must make links to the multiple named documents within a reference

Yes, and Jim's design meets those requirements.


More information about the rfc-interest mailing list