Cullen Jennings wrote:
If you want to get a kind such as the "software version" that is
specific to a node. Does that kind specify that a resource-id formed
from the node-id or is the fetch directly to the node-id.
To give another example of this, how should the diagnostic usage be
implemented? And how should similar future usages that define
peer-specific data work?
Some options:
1) use the current protocol to route a FETCH to a node-id rather than to
the resource-id. The message is delivered to the correct node and it
recognizes the kind (maybe there's a bit flag) that indicates this is a
node-specific kind rather than a standard distributed hash kind.
2) specify a new method with essentially the same semantics
3) specify a different method for calculating a resource-id for these
kinds such that the prefix for the resource-id for a node-specific kind
just happens to be the desired peer-id.
(I personally prefer option 1 because I think it's the simplest to
implement if you already have everything else.)
Bruce
_______________________________________________
P2PSIP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/p2psip
