Metadata on AbstractDataItems

[co-stig]

Metadata can be a JSON object, which we might want to attach to any data item object.

Example use cases:

• Integration with external systems like GitHub, e.g. storing repo name on a backlog and issue ID on a workitem
• Taking notes on workitems
• Recording pomodoro interruptions
• Grouping workitems by priority or importance
• Scheduling workitems
• More powerful tags

The simplest way would be to use PATCH semantics, i.e. using null values to delete keys. This means a single PatchMetadata strategy should be sufficient:

CreateWorkitem("66071b8e-ae7a-4e33-a51f-170881aa67e4", "3269262a-3f60-4f48-92fd-19949515945e", "Do something")
PatchMetadata("workitem", "66071b8e-ae7a-4e33-a51f-170881aa67e4", "{\"github-id\": \"101\"}")
PatchMetadata("workitem", "66071b8e-ae7a-4e33-a51f-170881aa67e4", "{\"notes\": \"See https://example.com\"}")
PatchMetadata("workitem", "66071b8e-ae7a-4e33-a51f-170881aa67e4", "{\"notes\": null}")
PatchMetadata("workitem", "66071b8e-ae7a-4e33-a51f-170881aa67e4", "{\"github-id\": null}")

The size of such JSONs shouldn't grow too much to stay efficient. For example, we probably shouldn't use it for implementing habit trackers, for example.