two separate issues, how to deal with IEEE Nan, Inf when you also have Maxima und, inf, minf. and how to get some requested precision in a result. The latter is not always computable. e.g. if the answer is zero but you fail to compute it, exactly.