Most errors in Fortran routines can be identified by the information
provided in Fortran runtime messages, which begin with the prefix
"FOR".
The Fortran compiler cannot identify all possible errors. The following
list identifies several errors not detected by the compiler that could
potentially result in problems:
- Failing to assign values to variables and arrays before using
them in your program.
- Specifying subscript values that are not within the bounds of
an array. If you assign data outside the array bounds, you can inadvertently
destroy data and instructions.
- Moving data into an item that is too small for it, resulting in
truncation.
- Making invalid data references to EQUIVALENCE items of differing
types (for example, integer or real).
- Transferring control into the range of a DO loop from outside
the range of the loop. The compiler issues a warning message for all
such branches if you specify OPT(2), OPT(3), or VECTOR.
- Using arithmetic variables and constants that are too small to
give the precision you need in the result. For example, to obtain
more than 6 decimal digits in floating-point results, you must use
double precision.
- Concatenating character strings in such a way that overlap can
occur.
- Trying to access services that are not available in the operating
system or hardware.
- Failing to resolve name conflicts between Fortran and C library
routines using the procedures described in z/OS Language Environment Programming Guide.