...
Early on in the design of Chameleon, we found the need to extend existing data types with additional fields. For example, each sport has unique data beyond team names, scores and status. While baseball might need information on who is up to bat, the count, team errors and so much more, a sport like hockey might need shots on goal, hits, penalty/power play information. There are 2 ways to approach this problem. Either add new fields to a relational data table (forever) or attach additional data to a core data type. We chose the 2nd method. All of the Chameleon data types have support for extending data by attaching dynamic tags to records. These dynamic tags can be text, numbers, dates, boolean, json or media. We even went all in on this idea by having a data type we call Custom which is essentially just dynamic tags.
...
And here are dynamic tags for an NHL player:
...
Note that we are working on a new data type: set. Basically, a set is a set of dynamic tags. This is a way for any dynamic tag to reference a set of dynamic tags providing full hierarchy of data. This could be used to store season or game information for players or teams. It could also be used to store different language versions for a story.
Custom Data Type
Not all data fits into standard Chameleon data types such as story, weather or finance. That’s where the Custom data type comes in. It is essentially key/value pairs attached to records. Chameleon calls these key/value pairs dynamic tags or dynamic fields. Although all the other data types have key/value pair support, the Custom data type is exclusively key/value pairs. Any 2 dimensional data is a good candidate for using the Custom data type. For example, a spreadsheet is a good example of what is a good fit for Custom where each column is a field and each row are the values to a record.
...