At 09:38 +0000 on 11/18/2013, David Stokes wrote about Re: Relative Branches / IBM macros:
Peter Relson wrote: For LLILF I would have said no since the LLILF operand is an "immediate". Well, I don't agree. Immediate operands are absolute expressions. Witness the fact that you get the "Symbol A undefined" error just the same writing say CLI label,A(123) (or AL1 for that matter), whereas LLILF clearly recognizes the symbol as being external/relocatable, but doesn't like that. It happily accepts say XX DS 0H ...... LLILF R15,XX-* (whatever sense that might make).
I think it is choking on the symbol being an external not just a relocatable. The latter it can compute. It might be interesting to see what happens if XX is in a different CSECT from the LLILF (ie: Will it accept it and adjust for the fact that the offset to the CSECT is unknown until Bind Time - you know the offset in the assembly but the Binder may move the 2 CSECTs apart).