// In Rust:
let mut my_tuple = (1, 2);
let first_elem = &my_tuple.0;

// error[E0506]: cannot assign to `my_tuple` because it is borrowed
my_tuple = (3, 4);
println!("{first_elem}");

Alas! That’s old news - the very first blog post on this site covered it. That post is frankly quite outdated now. It’s been years since then and Ante’s approach has substantially changed, having accrued many additions to make it more flexible. Now, it has changed again, having blended many elements from group borrowing with my own ideas to produce a more radically safe whole. In this post, I want to go over the new design and show how it leads to a more flexible kind of safe, shared mutability that is expressive enough to support cyclic mutability and complex data flow safely.

---

Why Shared Mutability?

Since many Rustaceans are skeptical of the value of shared mutability at all, it’s worth reiterating why being more lenient here is useful. More lenience means:

• Fewer borrow checker errors. Fewer errors both eases the learning curve for

  new users and speeds up iteration times for experienced ones.

• More design patterns are allowed. The observer pattern is difficult to get working without shared

  mutability. Cyclic types can also be more directly represented.

• Existing design patterns can have fewer workarounds. Fewer forced indices into vectors in particular.

  While this can be a useful pattern, I don’t believe the language should force users into this approach.

• More straightforward interop with other languages that also support shared mutability.

The last point in particular is more important than it seems. Ante is meant to be opt-in to ownership to some extent so that by default using it feels more like C# than it does Rust. This is an aspirational goal but allowing shared mutability greatly helps here.

---

Aside: What was wrong with Ante’s previous rules?

I imagine most will not be familiar with Ante’s old borrow checking rules, but for those curious I’ll lay out some more motivations for the change. If this isn’t you, feel free to skip this section.

I am not most readers

The original article mentions Ante’s ref and mut references as being &Cell<T> analogues in a way, and this has largely remained true at its core. Most of the extensions made since then were to extend this in some way, notably by introducing “local uniqueness” where a shared mutable reference only needed to be locally unique to be used where a uniq reference (think &mut T in Rust) was expected. This meant the compiler only needed to show that any variable used at the same time as a mut String did not mutably alias it. This ended up taking the form of a Distinct a b trait which was implemented if a did not occur anywhere within b. For example, if you use a Vec I32 at the same time a mut String is treated as a uniq String, the compiler would check a Distinct String (Vec I32) constraint in the background. Since a String can’t occur within a Vec I32, the code is allowed and the String is locally unique. This would be repeated for each variable used while the converted uniq String reference was still alive.

Allowing what is essentially a &Cell<T> to &mut T conversion like this was great for flexibility but had a few major flaws:

• Common types like String would occur within other types more often making them more difficult to use

  compared to less common types when converted into a local uniq reference.

• Distinct a b recurs into every field of b, including private fields. This meant adding a private

  field to a public struct in a library was a breaking change since it could make previously valid local unique conversions invalid.

• Converting between mut and local uniq happened often, and was confusing even though the compiler

  did it automatically.

• It relied on four reference types: ref and mut which allow shared mutability, and imm and uniq which don’t.

It wasn’t all bad though. Notably, it was powerful enough to encode safe, cyclic mutability, with the caveat that some Rc values had to be cloned (have their reference count incremented) during traversal.

For the reasons above though, it was clear the design had to change. The rest of the article is about the new design I eventually landed on which fixes all of these issues, but still keeps the good parts like safe, cyclic mutability2 and thread-safety.

---

The Core Problem

Why are some cases of shared mutability safe, like:

my_safe_fn (var pair: String, String) =
    name = ref pair.first
    pair.first := "Foo"
    println name  // Ok!

While others are not:

my_unsafe_fn (vec: mut Vec t) =
    elem = vec.get 1
    vec.clear ()
    println elem  // Dangling ref!

The difference is in what I call shape stability. If the shape of a type is stable, then there is no way to mutate it such that a previously valid reference into it becomes invalid. Picturing the references in memory helps here. Let’s look at how pair and name look in memory:

Before mutation:

      name
       |
       V
pair: ((first_ptr, first_len, ..), (second_ptr, second_len, ..))

After mutation:

      name
       |
       V
pair: ((first_ptr, first_len, ..), (second_ptr, second_len, ..))

Each String is an unboxed tuple of ptr, len, metadata, and each component of the outer pair is also unboxed. With this layout, there’s no way we could mutate pair to invalidate name!

Looking at vec and elem however, we see a clear difference:

Before mutation:

vec: (data_ptr, length, capacity)
          |
          V
          [0, 1, 2, 3]
              ^
              |
             elem

After mutation:

vec: (data_ptr, length, capacity)
          |
          V
          []
              ^
              |
             elem

elem is clearly dangling! The problem here is that mutating vec can mutate data_ptr, causing it to potentially point elsewhere when it is reallocated. Its shape has changed, leaving elem stranded.

It’d be nice if a reference type could remember whether it’s pointing at something stable or unstable - then we could do away with uniqueness in favor of a much more accurate restriction that rejects fewer valid programs.

---

Places Over Lifetimes

The first core part of the puzzle is the use of places instead of lifetimes.

Unlike a lifetime which is some arbitrary symbol representing an abstract region of code a variable or reference is valid for, a place is the exact set of things a reference may borrow from:

foo (x: I32) (y: I32): Unit =
    r1: ref 'x I32 = ref x
    r2: ref 'y I32 = ref y

    r3: ref '(x, y) I32 =
        if true then ref x else ref y

Above, r1 borrows from only x so it has the place 'x. Similarly, r2 borrows from only y so it has the place 'y. r3 may borrow from x or y so its place is the union of both: '(x, y). Places can also be paths such as 'z.bar to represent a reference borrowing from the struct field bar on a variable z.

Places don’t always name variables. There are anonymous places as well such as those for a reference to a temporary like ref 2, but that’s not important for this article.

Places make borrow checking easier in general too! Borrow checking in Ante is actually very simple: whenever a value is used, error if any place it refers to has already been moved. There is truthfully a little bit more to it than that, but that really is the core and it can be implemented in only a few lines of code.

---

Unstable Places

Once we have places we can model unstable places with a simple marker: !. An unstable place like a! will be invalidated by any mutation to the place a. This also means if we have 20 variables with place a!, we can mutate them all simultaneously since mutating them would only invalidate variables with place a!!.

This is a bit abstract though, so let’s see how this works in practice. Starting off, we have some Rust code which should be safe but gets rejected by the compiler:

fn rejected(v: &mut Vec<String>) {
    // error: cannot borrow `*v` as mutable more than once at a time
    let (first, last) = (v.first_mut().unwrap(), v.last_mut().unwrap());
    *first = "Mutate first".to_string();
    println!("{last}");
}

Rust’s compiler rejects this even though there is no way we can mutate first which would invalidate last. Theoretically, even if the first element is also the last, this code should still be valid3.

We can model this in Ante by flagging any references to returned elements of a vector as unstable:

// Declare a `first_mut` method taking a mutable vector and returning
// an unstable reference inside 'v to the first element
Vec.first_mut (v: mut Vec t): mut 'v! t can Fail = ...
Vec.last_mut  (v: mut Vec t): mut 'v! t can Fail = ...

Now, we can write the equivalent of rejected just fine:

not_rejected (v: mut Vec String) =
    first, last = v.first_mut (), v.last_mut ()
    first.* := "Mutate first"
    println last  // ok!

Above, first and last both have the type mut 'v! String so when we mutate first it would prevent us from using anything with a place of v!!4, but nothing does. So we’re free to use both first and last even if they alias.

Whether a reference is unstable or not is inferred in practice and doesn’t always need to be a manual choice by users. Any reference through a tagged union or element of a collection type will be unstable for example.

Loops

Loops and branching in general are fairly straightforward. They largely follow the same rules ownership has where an error is issued if a variable declared outside of the loop is invalidated within it:

var v = Vec.of [1, 2, 3]
one = v.get 0

for _ in 0 .. 3 do
    println one  // error: one may be invalidated by a later mutation in the for loop
    v.clear ()   // note: one invalidated here

Similarly, any closure which invalidates any part of its captured environment is treated as a FnOnce in Rust terms, preventing it from being called more than a single time.

Ante has effect handlers too! Control-flow for them is more complex and the rules for how they interact with invalidations of unstable places are documented here, but are subject to change (I think they’re a bit hacky). Note that handle interactions with unstable places are (at the time of writing) still unimplemented.

---

Exclusivity

It’s possible to write functions that are only valid if certain parameters don’t alias:

a_and_b_cant_alias (a: mut Vec String) (b: mut Vec String) =
    elem = a.first_mut ()
    b.clear ()
    println elem  // ok! But how?

In cases like these the compiler still needs some kind of exclusivity rule. In this case, when a parameter is mutated the compiler needs to ensure the caller of the function gives a unique value for it:

bad_caller (c: mut Vec String) =
    // error: 'a cannot alias 'b in this use of a_and_b_cant_alias, but 'c was used for both
    a_and_b_cant_alias c c

To check for this, whenever a mutation occurs the compiler issues a Mutate 'p effect which tags a function with the places that may be mutated. At call sites, each place that may be mutated must be distinct from other places instantiated5. Note that mut itself still allows arbitrary aliasing. It is only when an actual mutation occurs and Mutate 'p is issued does the compiler check call sites that any generic used doesn’t become 'p while it is still in scope.

Use vs Mention

Mutate 'p is an example of a use vs mention distinction. Code which merely mentions mutability isn’t subject to any uniqueness constraints. This is most notable since places can be paths so we may pass an entire mut Context to a function but it may only Mutate 'ctx.field. This scenario often occurs when iterating mutably over a field of some context value:

type Entity = name: String, health: I32
type Context = entities: Vec Entity, events: Vec String, score: U32

Context.on_death (ctx: mut Context) (entity: ref Entity): Unit
    can Mutate 'ctx.events
=
    ctx.score += 10
    ctx.events.push "${entity.name} died"

iterate_example (ctx: mut Context) =
    for entity in ctx.entities do
        entity.health -= 1
        if entity.health <= 0 then
            ctx.on_death entity  // ok!

In Rust the above code would be rejected, but there are actually several options in Ante. You can expose Mutate 'ctx.events on Context.on_death above, guaranteeing only that field may be mutated (ctx.score doesn’t need a similar annotation because mutating integers can never invalidate references), or you can tell on_death that entity may be stored within ctx to begin with:

Context.on_death (ctx: mut Context) (entity: ref 'ctx.entities! Entity): Unit
    can Mutate 'ctx
=
    ctx.score += 10
    ctx.events.push "${entity.name} died"

Above, we’re saying all of 'ctx may be mutated. For this specific example, this is less desirable since it’d mean entity as an unstable reference wouldn’t be valid after the call to on_death ends. For other patterns of code though, we may truly need to mutate all of 'ctx or keep it opaque6 and this would be exactly what we’d want.

No Need for Polonius

Taking a bit of a step back, I want to show a bit how powerful it is to reframe borrow checking in terms of places and mutations rather than lifetimes. Because lifetimes are tied to a region of code, Rust can struggle with complex data-flow:

fn get_default<'r, K: Hash + Eq + Copy, V: Default>(
    map: &'r mut HashMap<K, V>,
    key: K,
) -> &'r mut V {
    match map.get_mut(&key) {
    //    --- first mutable borrow occurs here
        Some(value) => value,
        None => {
            map.insert(key, V::default());
            //  ^~~~~~ error[E0499]: cannot borrow `*map` as mutable more than once at a time
            map.get_mut(&key).unwrap()
        }
    }
}

With Rust’s current model, getting the above code working requires more complex flow sensitivity (with the next-gen borrow checker called Polonius). In Ante, it just works:

get_default
    (map: mut 'r HashMap k v)
    (key: k)
    {Hash k} {Eq k} {Copy k} {Default v}
: mut 'r! v can Mutate 'r, Panic =
    match try do map.get_mut key
    | Some value -> value
    | None ->
        map.insert key (default ())  // ok!
        map.get_mut key ~> panic_on_fail

Ante’s borrow checker is based on places and thus doesn’t try tying references to any particular region of code. Instead, since the returned element reference from map.get_mut is unused in the None branch, it doesn’t need to be verified at all. This should make some intuitive sense. If the reference is never used, it shouldn’t matter whether it’s dangling or not.

Let’s look at another example:

struct Thing;

impl Thing {
    fn maybe_next(&mut self) -> Option<&mut Self> { None }
}

fn main() {
    let mut temp = &mut Thing;

    loop {
        match temp.maybe_next() {
        //    ^^^^ error: `*temp` was mutably borrowed here in the previous iteration
        //         of the loop first borrow used here, in later iteration of loop
            Some(v) => { temp = v; }
            None => { }
        }
    }
}

Above, Rust thinks the maybe_next borrow may last longer than the loop itself due to the temp = v; assignment to a variable defined outside of the loop.

type Thing = x: Unit

Thing.maybe_next (_: mut Thing): Maybe (mut Thing) = None

main () =
    var temp = mut Thing ()

    while true do
        match temp.maybe_next ()  // ok!
        | Some v -> temp := v
        | None -> ()

Meanwhile, the equivalent Ante code is okay since temp gets some anonymous place 'p, maybe_next returns the same 'p, and the mutation would only invalidate any unstable references to 'p.

---

Shared Mutability

We can use unstable places to safely allow mutating shared pointers like an Rc<T> (reference-counted pointer) in Rust.

The first realization is that since our analysis is based on places, Rc needs a place to represent the shared element: Rc 'p t. This isn’t needed on other pointer types like Box, but for Rc, this extra parameter 'p is necessary to link separate clones of the same value to the same internal value.

When modeling getting a mutable reference into the interior of an Rc, we need to make use of a union of places from earlier:

Rc.get (rc: mut 'outer (Rc 'inner t)): mut '(outer!, inner) t = ...

This type says that if we have a mutable reference to an Rc, we can get a mutable reference to the inner element. This returned reference will have two places: outer! (unstable because we’re crossing through a shared pointer boundary, and the Rc value itself may be swapped out), and inner - the shared value pointed to by every clone of this Rc.

Let’s try an example:

mutate_rc_bad (rc1: mut Rc 'p (Vec String)) (rc2: mut Rc 'p (Vec String)) =
    // string_ref: mut '(rc1!!, p!) String
    string_ref = rc1.get () |> Vec.first_mut

    // Mutates 'p, so string_ref with unstable 'p! is invalidated
    // note: string_ref may be invalidated by this mutation
    rc2.get () |> Vec.clear

    // error: string_ref may have been invalidated by an earlier mutation
    println string_ref

Great! We can safely mutate through an Rc value with no runtime checks, and the compiler still has our back if we try to do something unsafe. Is this enough to represent even cyclic references though?

---

Cyclic Mutability

For cyclic mutability we need… well, nothing else! We can just start using Rc to form cycles7.

type List 'p t =
   | Nil
   | Cons t (Rc 'p (List 'p t))

main () =
    var f = Rc.of (Cons false (Rc.of Nil))  // false -> []
    var t = Rc.of (Cons true f)             // true -> false -> []
    if f is Cons _ tail then
        // tail's places are '(f!!, a!), so it'd be invalidated by any mutation to f or the shared place 'a.
        // The latter 'a invalidation is important to prevent us from using it if we mutate any other shared node.
        tail := t

    emit_list t
        |> take 5
        |> iter println  // true, false, true, false, true

// The `do` here is unrelated, it just returns a thunk to be used as a generator function
emit_list (l: ref List t) = do
    if l is Cons elem tail then
        emit elem
        emit_list tail

This pattern is very similar to using GhostCell in Rust, but in a more ergonomic (no more GhostToken), built-in way. I’ll also emphasize that these checks are entirely static with no runtime checks.

If we wanted, we could even use shared mut type sugar to simplify things a bit. See the relevant section in the docs for that.

Note on Multi-threading

With more permissive shared mutability, how does Ante maintain thread-safety?

There’s not a very interesting answer here. Ante simply copies Rust’s Send/Sync traits. ref/mut references both allow for shared mutability and thus implement neither trait.

Instead, Ante has additional reference kinds imm and uniq which correspond to Rust’s & and &mut. These references are very rarely used - they’re more or less only used in multi-threaded code where after passing a thread boundary they can degrade back to ref/mut to call functions.

In the future, the kind of a reference will be attached to the place variable to more easily write code which is polymorphic over any reference kind.

---

Summary

This was a lot! Thanks to anyone sticking around to the end! It’s a lot to grok at once but despite the length this article ended up being, I do think the core concepts aren’t that complex. There are certainly rules to be had but most are in service of making the language more flexible and making more code “just work.” The main mental model is keeping track of the places a reference may borrow from, and whether any are unstable.

This article is mostly a restating of Ante’s shared mutability rules documentation, but I wanted to write an article version of it to follow up the original shared mutability article and to speak on some motivations behind the change.

If any of this interests you, join Ante’s Discord to keep up with the language and ask me questions! You can also clone the compiler at GitHub page to try out the language yourself. If you are interested in the project, the best way to help would be to help spread the project around to get some more visibility on it! Contributing would be appreciated as well of course (the compiler is small and I can help)!

---

1. Low-level is always a contentious term. I’m using it here to mean a language with unboxed

   types by default and no pervasive GC. If you have a preferred definition, let me know! ↩︎

2. The new design is actually more efficient for cyclic mutability! It doesn’t require the clones

   the previous design did. ↩︎

3. It wouldn’t be sound in Rust since multiple &mut references to the same

   data would violate aliasing assumptions, but it should be in our target semantics! ↩︎

4. ! on places can nest! 'foo! is an unstable place within 'foo while 'foo!!

   would be an unstable place within that previous unstable place. This could be, for example, a String element of a Vec (Vec String). If a type has multiple sources of instability (e.g. a struct holding multiple vectors), an API designer may choose to expose the separate fields, e.g. 'foo.bar! vs 'foo.baz!, or may choose to hide this by coercing to the parent place 'foo!. ↩︎

5. Any places that were already 'p are actually okay. The body of the function

   would already be aware of and check those cases. This enables patterns like passing in an immutable container and a mutable reference inside the same container. ↩︎

6. Future versions of Ante will allow for ‘abstract places’ where structs could

   declare a set of places inside of them to expose, separate from their actual fields. Library authors wanting to hide specific struct fields while keeping multiple places within the same struct distinct from each other in usual code could make use of this. The downside is that since these aren’t connected to specific struct fields, it’d be up to the library author to keep the translation to abstract places sound. ↩︎

7. Using Rc to form cycles still leads to leaks! Some other collection

   mechanism is needed to prevent leaks, but that is a separate issue from the mutability of a cycle. ↩︎