You can run a REXX program from the z/OS shells, or you can call it from any program just as you would call an executable program. The REXX program runs as a separate process; it does not run in a TSO/E address space.

A REXX program that is invoked from a z/OS shell or from a program must be a text file or a compiled REXX program that resides in the z/OS® UNIX file system. It must have read and execute access permissions. Each line in the text file must be terminated by a newline character and must not exceed 2048 characters. Lines are passed to the REXX interpreter as they are. Sequence numbers are not supported; if you are using the ISPF editor to create the REXX program, be sure to set NUMBER OFF.

If you are working in a z/OS shell environment and use only a filename to invoke the REXX program, the PATH environment variable is used to locate it. For example, myrexx uses PATH to locate the program, but ./myrexx searches only the working directory.

For a REXX program that is run from a z/OS shell or from a program, the SIR is established by default. If the REXX program calls a C program that is running POSIX(ON) or a program that requires an SIR, use the syscalls('SIGOFF') function to delete the SIR before calling that program.

CEXEC output from the REXX compiler is supported in the z/OS shell environments. To compile and put CEXEC output into the z/OS UNIX file system, you can use the REXXOEC cataloged procedure; it compiles under TSO/E and then uses the TSO/E OCOPY command to copy the compiled program from a data set to a file in the file hierarchy.