Hello, I’m struggling to understand how to use Base’s Maps. I found this bit of documentation which helped me figure out how to define my Map type: type rule = | Substr of string | Regex of string type account_rules = rule list Map.M(String).t But now I’m not sure how I can create a map sat…

It’s definitely true that the documentation around Map is insufficient. We’re working on it, and I also hope to refresh the chapter in RWO to cover the way Maps now work in Base (and will soon work in Core as well.) It’s definitely most convenient if you use Base’s Int module, but it’s easy enough …

[image] Yaron_Minsky: , because the functor mints a fresh type each time it’s called on a different input module, ensuring that different comparison functions have distinct type witnesses. This is a cool encoding, it may be worth a blog post : ) The only downside is that it is a bit user hea…

@bobzhang seeing you’re an Admin consider moving your more in-depth discussion into a thread with an appropriate title. (Discourse allows moving posts to a new thread.) I assume more people are interested in the more in-depth discussion you had with Yaron here but might not see it because it’s “hid…

Playing around with it a little, any idea why module M = struct include (val Base.Comparator.make ~compare:(-) ~sexp_of_t:(fun _ -> Sexp.Atom "dummy")) end let m = Base.Map.empty (module M) works but let m = Base.Map.empty (module struct include (val Base.Comparator.make ~compare:(-) ~sexp_of_t:(…

Because there is an existential type and the module must be bound (as in your first case) before we can access its components. You can find an explanation in this article page 7 in the section on Path Elaboration.

Wow, a great answer with an exact reference! Kudos. I guess first class modules are more counter-intuitive than I realized, since beta reduction doesn’t work as expected.

Modules have never been stable by substitution, first class modules don’t really change that. For example, in this, you can’t substitute M by its value (try to imagine what would be the type of a). module M : sig type t val x : t end = struct .... end let a = M.x

Fascinating. I guess it shouldn’t be surprising that the thing preventing module substitution is precisely that which separates modules from records, that being their contained types (which need static paths). Getting back to the topic of the new Base Map, here’s how I see it right now: Pros: No…

[image] bluddy: So you’re storing the comparison function in the data structure, which also has the added benefit of deliberately breaking polymorphic comparison. Neat. This actually reminds me a bit of Chris Okasaki’s approach in scads/design/heaps.md at e78233ac6a787b7c66b44cd6139392418b214…

It’s worth noting, the type discipline works without actually storing the comparison function in the data structure. See Map.Tree for a version of the data structure with the same type discipline, but no stored up function. It seems like a mistake to presume what Flambda would do with this kind of …

Examples of using Base Map

Learning

trefis February 16, 2018, 2:43pm 27

Note that the docs for v0.10 are now up, and that core_kernel’s doc for Map includes some examples: https://ocaml.janestreet.com/ocaml-core/latest/doc/core_kernel/Core_kernel/Map/

1 Like

Topic		Replies	Views
Functorial vs "comparator" container interfaces? Ecosystem core , base , core_kernel	12	2419	April 16, 2020
JaneStreet's OCaml Core library's Hashtbl interfaces Learning core , hashtbl	16	770	September 28, 2024
Base: design of empty in Map/Set/Hashtbl Learning	5	697	April 12, 2023
Functor trouble -- How to organize functor-heavy code? Learning functor	18	4108	December 14, 2021
Removing polymorphic compare from Core Ecosystem	20	6006	December 10, 2018

Examples of using Base Map

Related topics