Any sequence of multiple instructions is almost certain to be slower than the original CLC which I assume would be comparing with a literal, especially if the macro expansion required a branch. For baseless code, I normally still have a base register for a static area containing literals and constants, and I would expect that keeping the original CLC with a literal would be the simplest and fastest solution in any case where the operands would otherwise be the wrong way round.
Jonathan Scott, HLASM IBM Hursley, UK
