Dear Richard and all, I'd like to participate in the discussion and got some thoughts after reading the previous discussion and "draft-roome-alto-unified-props-new-00". Here they are:
IMHO Map and constraints solver (which contains linear/set constraints for ALTO, e.g, le, eq, subset, superset) are some kind of silver bullets for data-structure and algorithm respectively. Thus, about "the powerful filtered property map" Tony just mentioned, I prefer the solution a., which is similar to the general filter mentioned in https://mailarchive.ietf.org/arch/msg/alto/TYmEmha7SvMaquS_BhXTBcCmZ5c. In D3.3, per-query is definitely stateful, but per-resource may not. The per-query could be considered as per-resource by specifying property map (with different vtag) for each query. Determined by the implementation, it's possible to use a generic prop-map for every query which is stateless. In D4, I think the set of entities belongs to an aggregation could be considered as a kind of property. IP prefix implies a set of IPs in a compact way, but others may not. ALTO provide the dependency based on resource. The user can trace dependency on their on by querying such property for an aggregation (e.g., IP-prefixes for PID) and using it as a constraint in "the powerful filtered property map". Best Regards, Shenshen _______________________________________________ alto mailing list [email protected] https://www.ietf.org/mailman/listinfo/alto
