Re. SRFI 146 Draft #2 (2016-12-20).

  * map-set: The description indicates that later (in left-to-right
    order) associations for a key overwrite earlier ones but the test
    in the sample implementation (as well as the analogous set-adjoin
    of SRFI 113) use the opposite convention.

  * related to above: If the SRFI 113 convention is selected here then
    it may be worthwhile to consider alternate names for the map-set
    and map-set! procedures. (I feel that "-set" tends to be
    associated with overwriting the older value.)

Regards,

-chaw

The description was correct (should behave as hash-table-set); the implementation (and the test suite) was wrong.