Hi Andy,

I have a new version of the patch set which I believe addresses all of
your (quite reasonable) concerns.  The second patch changes the API of
the existing multi-valued operators as you suggest.  The third optimizes
the fraction case.  The fourth adds the new division operators.  The
fifth optimizes the existing truncate/round/floor/ceiling unary
operators.

The sixth patch removes a lot of code, by implementing division
operators in terms of other division operators.

The other notable patch is the last one, which is unrelated to numerics,
but it addresses a problem I discovered while implementing a complex
number class in GOOPS.  Suppose you define the following class within a
module:

  (define-class <gcomplex> ()
    (real-part #:getter real-part #:init-keyword #:real-part)
    (imag-part #:getter imag-part #:init-keyword #:imag-part))

There's no way I have found to make this work correctly.  If the module
exports real-part and imag-part, then those procedures will stop working
for normal complex numbers for those who import the module.  If the
module does not export them, then those who import the module won't be
able to use real-part or imag-part on <gcomplex> objects.  The last
patch fixes this.

    Best,
     Mark