+;;; should be used instead? (KLUDGED 2004-03-08 CSR, by replacing the
+;;; special variable references with (probably equally slow)
+;;; constructors)
+;;;
+;;; FIXME: As of 2004-05, when PFD noted that IMAGPART and COMPLEX
+;;; differ in their interpretations of the real line, IMAGPART was
+;;; patch, which without a certain amount of effort would have altered
+;;; all the branch cut treatment. Clients of these COMPLEX- routines
+;;; were patched to use explicit COMPLEX, rather than implicitly
+;;; passing in real numbers for treatment with IMAGPART, and these
+;;; COMPLEX- functions altered to require arguments of type COMPLEX;
+;;; however, someone needs to go back to Kahan for the definitive
+;;; answer for treatment of negative real floating point numbers and
+;;; branch cuts. If adjustment is needed, it is probably the removal
+;;; of explicit calls to COMPLEX in the clients of irrational
+;;; functions. -- a slightly bitter CSR, 2004-05-16