Generalized notion of auxiliary resources

[damooo]

Soild previously has a generalized concept of auxiliary resources for resources sharing the lifecycle of a primary resource. It helped to model supplementary resources like acls, metadata resources, etc in a general way, binding their lifecycle with primary resources.

Currently such general and extensible notion is not provided in lws spec. Instead metadata resources / linkset-resources alone have specialized lifecycle. Lifecycle of acls is not specified at all (When and How to create/update/delete them at all?).

Thus it helps to specify the general mechanism of auxiliary resources, with special description about linksets, and acls as their kinds.

Following are few reasons why it helps:

1. Unified specification for lifecycle of linksets, and acls, instead of ad-hoc handling each kind.

2. Ability to extend to have other kinds of control resources that controls schema, shape, content-exposition, profile, frame, and other details for each resource. These should have same lifetime as primary resource itself.

...etc.

[pchampin]

I agree that the spec should explicitly introduce a notion of "auxiliary resource". See my comment in #81.

As a consequence, I believe that the sections "Discovery" and "Container > Metadata" (or at least some parts of the latter) should be moved under a section

• describing the general concept of auxiliary resource, and
• describing the different types of auxiliary resources required from all implementations (currently storage description and link set, possibly others laters)

[acoburn]

This is being actively discussed in #165 (and related #167)