C445 Just one report here!

================================================================================

                             Referee Report

                         Complex Numbers for Java
                  by Michael Philippsen and Edwin Gunther

I. General Comments

   This is a solid contribution to the development of scientific computing
   capabilities in Java.  The techniques presented should be of wide
   interest.  Some elements of the presentation could be improved.
   I recommend acceptance after a revision.

II. Specific Comments

   a. Page 3, Section 3.1
      Change "... fields are no new keywords" to "... fields are not new
      keywords".

   b. Sections 3.2 and 3.3 are unintelligible unless one is very
      familiar with the gj system.  These should be rewritten to
      not have this dependency.  An appendix that relates this
      work to gj would be OK, or a little more of an introduction
      to gj would be better.

   c. Page 6, after bullet titled "Plain identifier"
      "mangeled" is a misspelling

   d. Page 6, in bullet "Selection"
      The description of the down-arrow superscript was lost on me on
      first reading.  This is something fairly subtle, but quite
      important.  Perhaps an example would help.

   e. Page 7, in bullet titled "Division"
      Change "For brevity, none" to "For brevity, neither"

   f. Page 8, first sentence
      I'm confused.  You claim that when both real and complex parts
      are stored together in a single double-sized array that there
      must be two boundary checks performed.  I am confused by this.
      I would think that only one check need be performed in this
      case.  Two boundary checks are necessary when there are two
      separate arrays.

   g. Page 8, sections titled "Array access as left hand side ..."
      and "Array access"
      By this point my head began to hurt.  I had a hard time
      disciphering the formal notation.  An example would help.
      Actually, examples would help in understanding all of these.

   h. Page 10, section 5.1
      You refer to a benchmark on "Microstrip calculations".  It is
      necessary to describe what this application is about and what
      its key numerical kernels correspond to.