The two sparse solvers provided by UmfPackSupport and SuperLUSupport deal differently with the input storage (at least if the input matrix is compressed and column major). UmfPackSupport only stores the pointers to the data while SuperLUSupport stores a copy of the data. Meaning that I can't really replace one by the other without either wasting memory or have some solver depend code. Did I miss something in the code?
Has this been done on purpose? I couldn't find anything about this in the documentation.