Floating point numbers

The Java floating point types float and double are represented in IEEE format, while the COBOL floating point types COMP-1 and COMP-2 are represented in /390 format. The conversion is performed automatically in the JCI functions.

During conversion, the following exceptional situations can occur, which are displayed to the caller as a condition variable in the field ResErrCode (COPY element JCI-METHODRES) when returning from the JCI function:

• COMP-1 ---> IEEE:

  RES-ERR-FLOAT-UNDERFLOW

  The /390 floating point number is lower than the smallest representable IEEE floating point number.

  RES-ERR-FLOAT-OVERFLOW

  The /390 floating point number is greater than the largest representable IEEE floating point number.

• COMP-2 ---> IEEE:

  (none)

• IEEE ---> COMP-1:

  RES-ERR-FLOAT-INVALID

  The IEEE floating point number equals NaN or infinity.

• IEEE ---> COMP-2:

  RES-ERR-FLOAT-UNDERFLOW

  The IEEE floating point number.is less than the smallest representable /390 floating point number.

  RES-ERR-FLOAT-OVERFLOW

  The IEEE floating point number.is greater than the largest representable /390 floating point number.

  RES-ERR-FLOAT-INVALID

  The IEEE floating point number equals NaN or infinity.

If bit positions are lost in the conversion, this does not lead to an exceptional situation.