On Wed, Mar 9, 2016 at 4:31 AM William D Clinger <xxxxxx@ccs.neu.edu> wrote:

Olin's SRFI 32 did not have that problem, because Olin required the
"linear update" procedures to return the sorted result regardless of
whether they had performed any side effects.

+1, it seems to be well-established since at least SRFI 1 that destructive operations return their result. It's trivial for the implementer to return the sorted list, and is often convenient for the caller. I think this alone is a compelling argument to have the destructive sorts return their result, without even getting into the performance argument.

Regards,

Kevin Wortman