Things are coming together... (yet another post about my Rete impl)

The Rete implementation I'm working on can now compile Conditions expressed as Python tuples to a Rete Network.

I'm using Robert B Doorenbos thesis 'Production Matching for Large Learning Systems' as a basis for my implementation but I've done away with several things that so far have seemed to be non-useful (Beta Memory objects among others). I'm counting on a few of those to come back and bite me further down the road.

To be able to debug the implementation I've also written a Rete Network Viewer which has proven real useful on more than one occasion. A, rather silly, Rule with the two Conditions ('?obj', 'n', '!=', 3) and ('?obj', 'n', '>', 2) produce the following Rete Network:



The green dots are the Alpha Network, the blue dots are the AlphaMemory objects, the Gray dots are JoinNodes, the Red dots are ConstantValueNodes and the Cyan dot is the ProductionNode.

Now *all* I need to do is implement removeFact, removeRule and a conversion function for turning Python objects into WMEs ;-) Piece of cake!