Pip Cet wrote:
> (nthcdr 1 (cons 'a 'b)) should probably be 'b, though.

Yes, Glenn also noted that bug. I fixed it by installing the attached patch. It 
also follows your suggestion of a quick path for small N (where we can omit both 
circularity and quit testing), plus to use EQ to compare objects. Thanks for 
checking the patch.