If getu(sys, sym::ArraySymbolic) with an ArraySymbolic, sym used to be collected. Now, it resolves the indices, idx = variable_indices(sys, sym) which, in my case, returns something like Int[1,2]. However this ends up as some _getp call
because sym_arr = [1,2] which is neither variable nor parameter of my inpr thus the timeseries idx do not contain ContinuousTimeseries().
Can be solved if get_all_timeseries_indexes(inpr, sym) returns Set([SII.ContinuousTimeseries()]) for random inputs (in this case: integers). But does this really make sense? For example, that's not how it is defined in SymbolCache, which just returns an empty set.
If
getu(sys, sym::ArraySymbolic)
with anArraySymbolic
,sym
used to be collected. Now, it resolves the indices,idx = variable_indices(sys, sym)
which, in my case, returns something likeInt[1,2]
. However this ends up as some_getp
callhttps://github.com/SciML/SymbolicIndexingInterface.jl/blob/82f1464e47c87ec402d0f368c091cb5232d9c07f/src/state_indexing.jl#L242-L245
because
sym_arr = [1,2]
which is neither variable nor parameter of my inpr thus the timeseries idx do not containContinuousTimeseries()
.Can be solved if
get_all_timeseries_indexes(inpr, sym)
returnsSet([SII.ContinuousTimeseries()])
for random inputs (in this case: integers). But does this really make sense? For example, that's not how it is defined inSymbolCache
, which just returns an empty set.https://github.com/SciML/SymbolicIndexingInterface.jl/blob/82f1464e47c87ec402d0f368c091cb5232d9c07f/src/symbol_cache.jl#L141-L149