filter: several issues
-
strip
does not take into account the hidden transitions (add tests about that)
- the code is fragile when we pass state numbers or transition numbers that do not exist: the dynamic_bitset overflows
- there is too much code duplication, in particular in the preparation of the dynamic bit sets: it's in the bridge, it's in a template function with quite an ugly documentation comment (TMP), etc. I believe that functions
hide_states
, hide_transitions
, parameterized on the container type, would do just fine. The user would do something like filter(aut).hide_states(ss).hide_transitions(ts)
.