support values with float suffixes e±N