Comparing fluents between states

[e-cal]

Each Step in a trace holds its State, which has its list of Fluent objects. Each Fluent object stores its value as a property.

This would mean the generators handle fluent values in one of two ways:

1. Each state has its own unique Fluent objects
   • Determining delta state would involve cross comparing all the fluent names and object lists to figure out which ones are actually the same
   • Seems very redundant
2. Each consecutive State overwrites the value from state to state
   • Only way to figure out delta state would be comparing what fluents are present and not present between 2 states (any fluents in both State's lists would have the same value regardless of actual state)

My idea would be instead to store a list of (Fluent, value) tuples on State and remove the value attribute from Fluent, or to store separate lists fluents and values that are indexed the same.

These changes would likely break how generation currently works. Any other ideas or workarounds?

[haz]

Hrmz...I think a State should have a mapping from Fluent to it's value. Then, the diff could be something akin to...

added = []
deleted = []
for f in state1:
    if state1[f] and not state2[f]:
        deleted.append(f)
    if not state1[f] and state2[f]:
        added.append(f)
return (added, deleted)

Means that Fluent needs a __hash__ function that's tied to its name. Something like..

def __hash__(self):
    return hash(str(self))