I think I have to retract what I said before.

The procedure `mapping-replace` from SRFI 146 should be directly
implementable through `mapping-search` as well. For this to work, we
need the full generality of `mapping-search`, which is currently in
the spec (this is what I missed).

So, now, I strongly vote in favor of John's (d) (for all relevant SRFI's).

Marc

Am Fr., 8. Jan. 2021 um 05:40 Uhr schrieb Arthur A. Gleckler
<xxxxxx@speechcode.com>:
>
> On Thu, Jan 7, 2021 at 1:42 PM John Cowan <xxxxxx@ccil.org> wrote:
>
>>
>> I agree.  So let's go with (c) for this SRFI, and for consistency insert a post-finalization note into SRFIs 113 (general sets and bags), 146 (mappings and hash maps), and 209 (enum-sets) something like this:
>>
>> "Implementations may choose to treat calling the update sub-procedure of {set,bag,mapping,hash-map,enum-set}-search as an error if the new key is not the same (in the sense of the comparator) as the original key.  This makes it possible to update the key as well as the value, but does not require changing the shape of the underlying data structure."
>
>
> Marc, do you agree with John about placing such a post-finalization note in SRFI 146?