bluebottlecoffee / tree-thinking Goto Github PK

the type and contingent max, min, step are unnecessary and sacrifice generality of the trees. It makes more sense for this data to not be present in the tree. Of course, they can be optional fields, but for the basic tree type, they're only misleading.

Rather than the current form of index and feature list, it makes more sense to just use feature keys. The sacrifice is that this requires there to be a hash of answers and these feature keys. However, this actually makes the validation between whatever produces your vectors, and the yaml_switch easier.

It would be helpful to have support for a tree version. I'll take ownership of adding support for a reader method to the ruby library.