TL;DR: The evaluation system uses eval() and evalChildren(). eval is supposed to be overridden by your own implementation. The evaluation logic uses Flags for marking the Nodes to be Dirty and/or Invalid.

Evaluation Functions

There are 2 main methods used for evaluation:

These functions are mutually exclusive. That means that evalChildren does not eval current Node, but only children of the current Node.

By default the implementation of eval() is “empty” and return 0. However it seems logical, that eval (if successfull) resets the Node not to be Dirty nor Invalid. This method is supposed to be overridden by your own implementation. As an example, you can check out the repository’s examples/example_calculator to have an inspiration how to setup the Node evaluation on your own.

The evaluation takes advantage of Node flags described below.

Node Flags

Each Node has 2 flags:

  • Dirty
  • Invalid

The Invalid flag has always higher priority. That means when the Node is Invalid it doesn’t matter if it is Dirty or not.

To mark a node Dirty or Invalid there are respective methods markDirty() and markInvalid(). Both methods take bool parameter for the new state. You can mark Node dirty by setting the parameter to True. Also you can un-mark the state by passing False value.

For both flags there are 3 methods available:

The same goes for the Dirty flag of course:

Descendants or Children are always connected to Output(s) of current Node.

When a node is marked Dirty or Invalid event methods onMarkedInvalid() onMarkedDirty() are being called. By default, these methods do nothing. But still they are implemented in case you would like to override them and use in you own evaluation system.