Updating enum_boxed_match according to audit.

[orizi]

Summary

Improved the implementation of enum_boxed_match libfunc to handle single-variant enums more efficiently. This PR fixes the code generation for boxed enum matching, particularly for the case of enums with a single variant where no branching is needed.

---

Type of change

Please check one:

• Bug fix (fixes incorrect behavior)
• New feature
• Performance improvement
• Documentation change with concrete technical impact
• Style, wording, formatting, or typo-only change

---

Why is this change needed?

The current implementation of enum_boxed_match was not correctly handling single-variant enums. It was generating unnecessary code for the single-variant case, which should be a simple pass-through operation without branching logic.

---

What was the behavior or documentation before?

Previously, the enum_boxed_match libfunc was generating unnecessary code for single-variant enums, treating them similarly to multi-variant enums which require branching logic.

---

What is the behavior or documentation after?

Now, for single-variant enums, the enum_boxed_match libfunc generates simpler, more efficient code that directly returns the variant box without any branching. The implementation also includes better handling of offsets and cleaner code for the multi-variant cases.

---

Additional context

Added several test cases to verify the correct behavior for single-variant and multi-variant boxed enum matching. The changes also include some refactoring to improve code clarity and maintainability, particularly in the utility functions for type handling.

[reviewable-StarkWare]

This change is 

[orizi]

• Updating enum_boxed_match according to audit. #9600 👈 (View in Graphite)
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 6e9942c64c

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[orizi]

@orizi made 1 comment.
Reviewable status: 0 of 7 files reviewed, 1 unresolved discussion (waiting on @ilyalesokhin-starkware and @liorgold2).

[orizi]

@orizi resolved 1 discussion.
Reviewable status: 0 of 7 files reviewed, all discussions resolved (waiting on @ilyalesokhin-starkware and @liorgold2).

[ilyalesokhin-starkware]

crates/cairo-lang-sierra-to-casm/src/invocations/enm.rs line 471 at r3 (raw file):

    }];

    assert!(boxed_enum_ptr.apply_known_ap_change(1));

Suggestion:

// Fix boxed_enum_ptr to account for the `inc_ap` above.
assert!(boxed_enum_ptr.apply_known_ap_change(1));

[ilyalesokhin-starkware]

tests/e2e_test_data/libfuncs/enum line 770 at r3 (raw file):

}

extern fn enum_boxed_match<T>(e: Box<T>) -> BoxedSingle nopanic;

this also works, right?

Suggestion:

Box<felt252>

[ilyalesokhin-starkware]

@ilyalesokhin-starkware reviewed 8 files and all commit messages.
Reviewable status: all files reviewed, 2 unresolved discussions (waiting on @liorgold2 and @orizi).

[orizi]

@orizi made 2 comments.
Reviewable status: all files reviewed, 2 unresolved discussions (waiting on @ilyalesokhin-starkware and @liorgold2).

---

tests/e2e_test_data/libfuncs/enum line 770 at r3 (raw file):

Previously, ilyalesokhin-starkware wrote…

this also works, right?

yes it does.

---

crates/cairo-lang-sierra-to-casm/src/invocations/enm.rs line 471 at r3 (raw file):

    }];

    assert!(boxed_enum_ptr.apply_known_ap_change(1));

Done.

[ilyalesokhin-starkware]

tests/e2e_test_data/libfuncs/enum line 770 at r3 (raw file):

Previously, orizi wrote…

yes it does.

Lior said he prefers that syntax.

[orizi]

@orizi made 1 comment.
Reviewable status: 7 of 8 files reviewed, 2 unresolved discussions (waiting on @ilyalesokhin-starkware and @liorgold2).

---

tests/e2e_test_data/libfuncs/enum line 770 at r3 (raw file):

Previously, ilyalesokhin-starkware wrote…

Lior said he prefers that syntax.

Done.

[ilyalesokhin-starkware]

tests/e2e_test_data/libfuncs/enum line 862 at r1 (raw file):

branch_align() -> ();
enum_boxed_match<test::A>([1]) -> ([2]);
branch_align() -> ();

Why did the branch_align disappear?

Code quote:

branch_align() -> ();

[orizi]

@orizi made 1 comment.
Reviewable status: 6 of 8 files reviewed, 3 unresolved discussions (waiting on @ilyalesokhin-starkware and @liorgold2).

---

tests/e2e_test_data/libfuncs/enum line 862 at r1 (raw file):

Previously, ilyalesokhin-starkware wrote…

Why did the branch_align disappear?

lior's suggested change made this function not be considered as a branch on sierra-gen - so no branch-align was caused.

in general we may want to consider this for "single-branch-matches" but not for this PR.

[orizi]

@orizi made 1 comment.
Reviewable status: 6 of 8 files reviewed, 3 unresolved discussions (waiting on @ilyalesokhin-starkware and @liorgold2).

---

tests/e2e_test_data/libfuncs/enum line 862 at r1 (raw file):

Previously, orizi wrote…

lior's suggested change made this function not be considered as a branch on sierra-gen - so no branch-align was caused.

in general we may want to consider this for "single-branch-matches" but not for this PR.

fixed underlying issue in another PR.

[liorgold2]

@liorgold2 reviewed 3 files and all commit messages, and made 1 comment.
Reviewable status: 5 of 8 files reviewed, 4 unresolved discussions (waiting on @ilyalesokhin-starkware and @orizi).

---

tests/e2e_test_data/libfuncs/enum line 812 at r8 (raw file):

enum B {
    Value1: A,
    Value2: A,

Actually, let's have the second item larger, which will put the inner selectors in different locations due to padding. E.g.

enum B {
Value1: A,
Value2: (A, felt252, felt252)

[liorgold2]

@liorgold2 reviewed 3 files and made 3 comments.
Reviewable status: 6 of 8 files reviewed, 7 unresolved discussions (waiting on @ilyalesokhin-starkware and @orizi).

---

crates/cairo-lang-sierra-gas/src/core_libfunc_cost_base.rs line 412 at r8 (raw file):

                // plus the regular match cost - but only when branching is actually needed.
                let n = signature.branch_signatures.len();
                match n {

Suggestion:

                match signature.branch_signatures.len() {

---

crates/cairo-lang-sierra/src/extensions/modules/enm.rs line 397 at r8 (raw file):

    ) -> Result<(Vec<ConcreteTypeId>, bool), SpecializationError> {
        let type_info = context.get_type_info(ty)?;
        let (inner_ty, is_snapshot) = peel_snapshot(ty, type_info)?;

There was a question whether it's ok to peel at most one snapshot. What happens with Box<@@T>?

---

crates/cairo-lang-sierra/src/extensions/modules/enm.rs line 461 at r8 (raw file):

            context,
            enum_type.clone(),
            variants.iter().cloned(),

Is this efficient? Does it create an interator that clones all items when necessary?

Code quote:

         variants.iter().cloned(),

[orizi]

@orizi made 4 comments.
Reviewable status: 6 of 8 files reviewed, 7 unresolved discussions (waiting on @ilyalesokhin-starkware and @liorgold2).

---

crates/cairo-lang-sierra/src/extensions/modules/enm.rs line 397 at r8 (raw file):

Previously, liorgold2 wrote…

There was a question whether it's ok to peel at most one snapshot. What happens with Box<@@T>?

there's no such thing at the sierra level - there's either one or none.

---

crates/cairo-lang-sierra/src/extensions/modules/enm.rs line 461 at r8 (raw file):

Previously, liorgold2 wrote…

Is this efficient? Does it create an interator that clones all items when necessary?

it only clones on "next" calls - no cloning of the vector is performed.

---

tests/e2e_test_data/libfuncs/enum line 812 at r8 (raw file):

Previously, liorgold2 wrote…

Actually, let's have the second item larger, which will put the inner selectors in different locations due to padding. E.g.

enum B {
Value1: A,
Value2: (A, felt252, felt252)

Done.

---

crates/cairo-lang-sierra-gas/src/core_libfunc_cost_base.rs line 412 at r8 (raw file):

                // plus the regular match cost - but only when branching is actually needed.
                let n = signature.branch_signatures.len();
                match n {

Done.

[ilyalesokhin-starkware]

@ilyalesokhin-starkware partially reviewed 3 files and resolved 3 discussions.
Reviewable status: 7 of 8 files reviewed, 4 unresolved discussions (waiting on @liorgold2).

[liorgold2]

@liorgold2 partially reviewed 6 files and all commit messages, made 1 comment, and resolved 4 discussions.
Reviewable status: 7 of 8 files reviewed, 1 unresolved discussion (waiting on @orizi).

---

crates/cairo-lang-sierra-to-casm/src/invocations/enm.rs line 444 at r9 (raw file):

    if num_branches == 1 {
        // The box should point to: base_addr + 1 (selector).
        let variant_box = CellExpression::add_with_const(boxed_enum_ptr, orig_offset + 1);

Consider checked\_add or explicitly checking orig_offset < ... before.

Code quote:

orig_offset + 1

[liorgold2]

@liorgold2 reviewed 3 files.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @orizi).

[liorgold2]

(see comment below)

@liorgold2 made 1 comment.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @orizi).