But you can specify in ML if the posted from data is written as numeric or string. Why not add an additional option that lets you chose between string1 (within the raw data) and string2 (into an extra file) (and even better: string3 using both)?