Ric rightly writes:

> >Although the manuals may suggest an interpretation something like the
> comment
>  >above, your statement is a mild misinterpretation of what is actually
being
>  >said. The way that the manual's comments should be read is: if you need a
>  >COBOL binary integer between 5 and 9 digits, then a J2 datatype will fill
>  >your bill. A J1 datatype won't be big enough, and a J4 would be overkill.
>  >
>  >What's not being said, even though it could be read that way, is that a J2
>  >datatype can only legitimately contain numbers with a minimum of 5 digits
> and
>  >a maximum of 9 digits. A COBOL-legitimate datavalue for a J2 field can
> range
>  >from 0 (1 digit) to plus or minus 999999999 (nine digits) -- and that's
> only
>  >if the application program enforces those rules. IMAGE itself, as I
> mentioned
>  >earlier, doesn't impose any restrictions on the bit patterns that can be
> put
>  >into the 32-bits of a J2 dataitem.
>  >
>  >Wirt Atmar
>  >
>
>  I disagree.  I have never heard of a simple data type that doesn't allow
>  the value of 1, 2 ...  What the manuals are stating is that if you will
>  need from 5 to 9 decimal digits, then it will take a 32 bit binary field to
>  store that range of numbers (J2).  That doesn't exclude the values from 1
>  to 4 digits.  (Of course it's all base on the fact that decimal numbers
>  don't map directly into binary numbers.)  In other words, the value 1 is a
>  legitimate value for a J2 datatype in Image just as it is in a Cobol S9(9)
>  Comp field.

No, you don't disagree. You simply restated exactly the same thing I tried to
say above.

What I disagreed with -- and so did you -- was the earlier sentiment that a
J2 datatype is constrained to a minimum of 5 digits and a maximum of 9. The
latter condition is true (sort of, if the application enforces it). The
former is not.

Wirt Atmar