IBM Mainframe Forum

I'm getting this error when entering a number that is too big for the field (for example, if I enter 999 for a N1.4 field).

I would like to catch this situation and display a more meaningful error message if possible. Is it possible to bypass the Natural system error and display my own error message?

I tried it using a few variations of ON-ERROR / IF *ERROR-NR = 1125, REINPUT " " , but it still is displaying the system error.

I'm thinking it's possible to work with the field as an alpha instead, but I would like to avoid having to do that if possible.

Yes, we all fight this one.

Natural sends the message due to an error condition, and he does not relinquish control (back to your program) until the error is resolved. So the ON ERROR and IF *ERROR-NR statements won't execute until the user fixes the problem - too late.

Change the input field from numeric to alpha. Then decide if you want to make things easy for yourself or for your user. I recommend the latter.

If the user knows that the input must be entered in the format number-decimal-number-number-number-number, then you can use the MASK clause.But this doesn't allow flexibility for input values such as "1.1" or ".123", nor does it allow for negative values.

For flexibility, use the VAL function. But now you need to be concerned with too many decimal digits. That is, the user could enter ".12345" and VAL will truncate.Note that the input value is tested for non-blank, as older versions of Natural will abend when VAL is presented with a blank.

Thanks for the helpful info

I also learned something new about VAL and FRAC (which I never have seen used before), and the ability to use an expression like IF #A IS (N1.4)

Thanks again!

How do we easily identify the errant field if there are a lot of them on the stack?

Not by reanimating a five-year-old thread.