There are several issues with transition systems which need to be fixed. Most of them are caused by introduction of regulations, which in consequence caused that a state is not identified just by its content, but also by the sequence of rules used (in general).
[ ] By saving TS, information about uniqueness of states is lost - the sequence information is not stored (and its probably not desired to change that).
[ ] Loading of TS does not consider that some states might be duplicated, and some states might be lost in the process.
[ ] Encoding/Decoding of states is not reversible anymore in general.
[ ] There is a mess (although already improved) in types of states w.r.t. content (multiset vs. vector) ad history length (memoryless, onestep, "infinite").
[ ] Direct TS (for multiset type of states) should be merged with TS for vectors.
[ ] TS comparison (__eq__) is not robust and final step (comparison of edges) behaves very strange (might be an issue with hashing).
There are several issues with transition systems which need to be fixed. Most of them are caused by introduction of regulations, which in consequence caused that a state is not identified just by its content, but also by the sequence of rules used (in general).
__eq__
) is not robust and final step (comparison of edges) behaves very strange (might be an issue with hashing).