RFC PR: Add `fmap` as a `map` alternative with a better specification

CameronBieganek · July 29, 2026, 6:53pm

I’m looking for feedback on the following RFC PR, which proposes a new fmap function with a stricter API than that provided by map.

github.com/JuliaLang/julia

[RFC] Add `fmap` as a `map` alternative with a better specification (#62518)

master ← CameronBieganek:clb/add-fmap

drafted 10:21PM - 26 Jul 26 UTC

CameronBieganek

+71 -0

This is a proposal for a new mapping function called `fmap` with a more precise …specification than `map`. Of course the `map` function is not removed or deprecated by this PR, but future code would be encouraged to use either `fmap` or `Iterators.map`. The `fmap` function is inspired by `fmap` from Haskell. The output type of `fmap` is always "similar" to the input type, where "similar" means the container type is the same but the element type might be different. The `map` function does not always follow that rule. I would like to develop a consensus on this addition and on the details of the docstring before I move ahead with implementations, so this draft PR currently only includes the docstring for `fmap`.

If possible, please add your comments to the PR rather than here on Discourse.

Topic		Replies	Views
Inconsistency between map and fmap General Usage question , functors	2	348	November 14, 2023
In-place map General Usage	3	1635	September 29, 2020
Why `map` requires `collect` to use for general iterable things? General Usage question , iterators , map	18	799	May 28, 2024
Map as a partial function Internals & Design	4	1839	June 29, 2018
Define map, filter, ... also as functionals Internals & Design	10	1691	September 19, 2022

RFC PR: Add `fmap` as a `map` alternative with a better specification

Related topics