Should Directories store TD Models as well as Instances?

w3c / wot-discovery

Repository for WoT discovery discussion

https://w3c.github.io/wot-discovery/

Other

19 stars 17 forks source link

Should Directories store TD Models as well as Instances? #58

Open mmccool opened 4 years ago

mmccool commented 4 years ago

For developer time use case (see https://github.com/w3c/wot-usecases/pull/33)
Would also like a "link" from Instances to Models
Models need to live somewhere with stable URLs
Convenient if they are in the same directory service as TD instances, with the same API, query support, etc.
Could still store instances and models in different databases in practice
Complications: directories needs to know if a particular TD-like-file is a TD or a TDM so it can apply the correct validation procedure

mmccool commented 4 years ago

To capture:

npm comparison
Digital twin use cases
Vorto use case (stores models)
ODM connection/components (convert to TDMs?)
Normative "Link" relation types in TDs to TDMs ("implements") and maybe ODM capabilities (or TDMs representing those capabilities)
probably want the ability to discover TD instances that "implement" certain TDMs

To elaborate:

Use case workflows, esp for development
How does this relate to profiles?

relu91 commented 3 years ago

npm comparison

If we use npm as a similar concept, Thing Models are not needed at run time. Developers will download TMs and use the downloaded document at run time (i.e. create TDs, Discovery other TDs, etc.). Just like npm has npm install. I really don't see a strict requirement to have TM in a TDD, probably it would be better to define a different service (i.e. a Thing Model Directory).

memelet commented 3 years ago

@mmccool Would you elaborate the context for “digital twin”? In general I’m struggling to see how wot distinguishes between the actual thing and it’s twin.

... After searching and reading thru the wot projects, it seems that a TD with bindings would be defined for the physical device (akin to an edgex device service or hono’s concept of a device), and a separate TD with probably different bindings would be defined for the twin (akin to ditto’s concept of a thing).

In short, wot seems to not model concept of device vs device-twin in any concrete way. Is this understanding correct?

relu91 commented 3 years ago

@mmccool Would you elaborate the context for “digital twin”? In general I’m struggling to see how wot distinguishes between the actual thing and it’s twin.

... After searching and reading thru the wot projects, it seems that a TD with bindings would be defined for the physical device (akin to an edgex device service or hono’s concept of a device), and a separate TD with probably different bindings would be defined for the twin (akin to ditto’s concept of a thing).

In short, wot seems to not model concept of device vs device-twin in any concrete way. Is this understanding correct?

Thank you for your feedback. I think it might be worth opening a new issue in https://github.com/w3c/wot-architecture or https://github.com/w3c/wot since you're asking broader questions than the scope of this issue. Also there you'll reach more people.

mmccool commented 3 years ago

I agree, this needs to be discussed in architecture or thing-description.

mmccool commented 2 years ago

Resolved as "no" for now, but we might consider in next version, so I have marked it as deferred.