Tony, You were too polite to say so, but the current rules also make little sense for SETAF and SETCF.
SETCF is, however, usable. In order, for example, to obtain the absolute value of a complex number, ABS(a + bi), for some table-generation applications I wrote a SETCF routine in PL/I that accepted two external engineering-notation character strings, signed a and b, converted them into internal DFP values, implicitly calculated D = +sqrt(a**2 +b**2) using the generic PL/I bif ABS, converted D back into an external engineering-notation character string, and returned it to the assembler. This is, I suppose, baroque; but it is much better than trying to do floating-point arithmetic interpretively in the macro language. (I have worked out how to normalize and compare/sort engineering-notation values in the macro language, but I have balked at writing macros to obtain the square root of one or to multiply a pair of them together.) John Gilmore, Ashland, MA 01721 - USA
