initial implemenation of feature?

[dcj]

Here is one potential implementation of a feature? predicate.
Obviously it assumes the current feature implementation/representation as a hashmap.
Perhaps there is a better test for feature than I have proposed here....

[willcohen]

Based on (#2 (comment)), #2 may need to get settled before we figure #3 out. My initial hunch -- and it might be too far on the side of maintaining interoperability -- is that ovid might want to have multiple ways to represent a feature.

My biggest question is around tech.ml. I still don't totally understand how these featurelike rows will interact with those except that there's some way that metadata will likely come into play. If it's as simple as metadata clarifying something, then maybe those features just tick one of the boxes on ovid.

With that in mind, maybe aurelius becomes a little more opinionated. Ovid LETS features interoperate with one another no matter their representation (much like geo does with JVM types), but aurelius recommends a clean way to go?