[m-dev.] map.map_values_foldl

[Julien Fischer]

On Mon, 10 Jun 2024 at 22:40, Zoltan Somogyi <zoltan.somogyi at runbox.com> wrote:
>
> There is a bit of an inconsistency between map.map_values
> and map.map_values_foldl. The predicate that expected to be passed
> to map.map_values has the signature pred(K, V, W), where K is the key type
> of the map, and V and W are the value types in the initial and final maps
> respectively. There is a separate map.map_values_only predicate
> which expects a predicate with the signature pred(V, W). So the
> predicate being mapped
>
> - has a K parameter in map.map_values, but
> - has no K parameter in map.map_values_only.
>
> That works. Sometimes you need the K parameter, and sometimes
> you don't, in which case having to add an always-ignored argument
> to the predicate you would pass to map.map_values would be
> an annoyance.
>
> The problem is that map.map_values_foldl, and map.map_values_foldlN
> for N=2, 3 do not follow this convention. They all take a predicate
> whose signature lacks the K paramater. This makes the naming scheme
> inconsistent.
>
> The obvious fix would be to rename those three predicates by doing
> s/values/values_only/, and reuse their old names for versions that
> expect predicates that do take a K parameter. This would be a breaking
> change, though the break would be trivially fixable by a rename.
>
> Would anyone object if I made that change?

No, go ahead.

Julien.