yes. m/relieve doesn't accumulate and doesn't change payload type, it's only about consumption rate. (m/relieve conj >xs) is likely wrong because conj argument types are incompatible

m/relieve function signature should be T->T->T

Ok, m/relieve docs say "`When upstream is faster than downstream, overflowed values are successively reduced with given function rf.`

Is the doc correct or am I missing some nuance

the doc could be better

the values are reduced together not against an accumulator

ok so + works but conj is a type error

because (conj 1 2) on the first run

it would produce nested vectors, probably not what you want

on numbers it fails due to type error yes