Plug-ins will enable external developers to extend the core playground, making sure they can implement their own use cases. The primary use for plug-ins is in the prototype runtime. This means that prototype developers can easily re-use plug-ins in their own Python code.
Basics
Logically, there will be 2 main types of plug-ins:
- VSS plug-in
- Implements / overwrites a VSS Python class
- Can also do visualization via JavaScript
- Interacts with backend via VSS data broker, if needed
- Examples: “Moving vehicle plug-in”, includes
- JavaScript impl. of Google-Map like visualization
- A set of VSS interface implementations to read the vehicle speed, acceleration, etc.
- General plug-ins
- Does not implement a specific VSS interface
- Communicates with a generic cloud service (not via data broker)
- Examples: “AI inference for Smart Wipers”
There will be 2 levels of visibility for plug-ins
- Protected
- Visible / editable for invited users
- Public
- Visible for all in the “plug-in web store”-list of plug-ins
- Editable for invited users
Plug-ins will have meta-data
- An image, e.g. used on the plug-in home page
- An icon, e.g. used in the "Analysis" tab
- Config data for remove web services (e.g. as an editable JSon file...?)
Backend Integration
- Plug-ins can call any remote web service
- We might want to consider providing a easy way of encapsulating this, at least for standard GET functions?
- Limit to synch calls only initially OK
- Have to ensure in particular that we can contact a VSS Data Broker / Kuksa instance