GAI main classes - makingthematrix/gailibrary GitHub Wiki

To access GAI, the developer will have to create an object (or maybe access a singleton - not decided yet) of the GAI struct:

struct GAI {
  cells: CellRegistry
  cellTypes: CellTypeRegistry
  functions: FunctionRegistry
  graphs: GraphRegistry
  ids: Identifiers
  conf: Configuration
}

All the names may still change, especially these *Registry names. I've read once that if a programmer does not know what the object really does, she names it a "manager". It's a bit like this here. The main object, GAI, wraps together "registries", ie. maps of CellId to Cell, CellTypeId to CellType, FunctionId to Function, and GraphId to Graph. Each registry will also provide utility methods specific for the given type. On top of that, Identifiers can be used to generate and maintain ids used in these four, and Configuration will hold configuration data. Plus, GAI itself will provide methods to setup everything, run iterations, and collect results.

This might be a good moment to stop and think about what form of Dependency Injection I want to use. It's a big subject, but one I prefer to leave for later. I have a feeling that in this case making a decision too early might result in that, as the library grows, I will be forced to use a solution which doesn't really fit the architecture. So, I think that while GAI is small I can live without deciding on a specific solution. I will come back to it when the first working version of GAI is ready.

Apart from the GAI data structure, the library will provide a few utility classes with generic functions, function builders, operations on sets, etc. - anything that may make the programmer's work easier. I hope that in the future I will be able to expand these classes with new functions written by others, or by me but because of the feedback I'll get from people using GAI.

Identifiers <- | -> The main loop