The back-end right now assumes SAFA AND nodes only have epsilon transitions.
That is not true; AND nodes only have skip transitions and it's always the case that those skips are total and disjoint:
Union_n [s1, s2, ... sn] = [0, infty)
Insersection_n [s1, s2, ..., sn] = {}
This assumption is crucial for negation to work correctly and we now violate it by having nodes [e1 = \epsilon, e_2 = \epsilon, ...., e_n = \epsilon] such that
Union_n [s1, s2, ... sn] = \epsilon, not the full interval (not total).
The back-end right now assumes SAFA AND nodes only have epsilon transitions.
That is not true; AND nodes only have skip transitions and it's always the case that those skips are total and disjoint:
This assumption is crucial for negation to work correctly and we now violate it by having nodes
[e1 = \epsilon, e_2 = \epsilon, ...., e_n = \epsilon]such thatUnion_n [s1, s2, ... sn] = \epsilon, not the full interval (not total).