Comment made by: gshayban
Marshall, there really isn't a proposed enhancement, yet. So there's nothing to be against! Your input is valuable. (Regarding c.c.reducers, that is a separate problem -- yes that behavior is surprising)
Considering kv-support for transducers:
Is it useful to have some functions that transform reduce-kv style reducing functions (fn (link: result k v))?
filter- * These could be mechanically generated. You wouldn't have to have a kv-version for every transducer currently in core. Some like map or filter could specifically apply to the key and ignore the val, or v.v.
Some things like map's transducer would be arity-incompatible (map's transducer has a varargs arity).