> Note, by the way that while my assq is arguably uglier than yours,
       > mine is O(1) space and yours is O(N) where N is the length of the
       > alist.

D'oh!  Not really -- I just figured out what `mn_to_cdr' must mean.
Sorry.

Still, it's noteworthy that to achieve O(1)ness you had to write in
linear style not procedural style -- the variable `pair_key' in
particular -- and that you had free two references in a loop rather
than unprotecting a frame just once, outside the loop.

-t