by cgrand (🦋 🦣 𝕏) There is a modeling technique I’ve used in the past, and the more I use it, the more it feels like the right default in a certain class of...
  • Simon Brooke@mastodon.scot
    1·
    1 month ago

    @yogthos The inference engine I was working on for my thesis (and never completed), Wildwood, assigned objects to classes automatically based on their features; so that

    1. one object could be in many otherwise unrelated classes; and
    2. you could define a new class, and it would be automatically populated with every object already known to the system which met its criteria.

    I wrote about this idea in my first essay of the #PostScarcitySoftware project.

    https://www.journeyman.cc/blog/posts-output/2006-02-20-postscarcity-software/#social-mobility

    • ☆ Yσɠƚԋσʂ ☆@lemmy.mlOPM
      1·
      1 month ago

      Context dependent types is a really good concept in my opinion. Having a small set of common basic types that are shared across the language is very convenient, but enriching these types within a specific domain can provide a lot of value by adding metadata on top. As a bit of data naturally evolves and acquires new properties over its lifecycle, there needs to be a way to dynamically migrate to whatever structural category actually fits its current shape. This sort of stuff becomes particularly important when dealing with authentication or validation rules where we might want to enrich the data with entirely new attributes.

      And I also agree that a lot of the decisions about the way languages work stem from the constraints that were in place in the early days of computing, but are no longer there today. I really like the framing of post scarcity computing.