>Why you don't want to keep the field NULL ?

It's been a policy since years. NULL is a cause of problems and we ("I") always got rid of that. Whenever a record is created, all fields have a default value, empty that is. That goes for Character, Numeric, Boolean, Text, Date, etc. A good thing following that path is that if we see a NULL somewhere, this is an indication that something went wrong and we can correct it. Otherwise, we wouldn't have any indication of it in regards to the comparison as to know if it is normal or not. Also, when scanning a dataset, we can talk the same language, which means we wouldn't have to code our code to check if the field is null before trying to make a calculation with it, etc. Those are just a few adavantages I can see at first. But, I know, this has saved me hours and hours since years and we collected great success with that approach.