>> QDOSMSQ's main problem , as you have found, is not so much the range of > > numbers that can be represented, but the way that the software displays it. > > I'm sure some of the procedure wriuters amongst us, who understand these > > floating point things (I don't !), could write a routine to take in a float > > and return a 'proper' string representation for display purposes ........... Actually, this isn't so easy as it appears, since you would most likely do that by successive substractions - and that will then introduce quite some rounding errros! What would be needed would be another way to represent/store the fp numbers entirely. (and now, I presume Marcel will go and prove me wrong... :-) )
Wolfgang
