[Bug] Missing derive all for events

[Jinwei1987]

Component

contract

What version of Alloy are you on?

1.8.3

Operating System

macOS (Apple Silicon)

Describe the bug

    alloy::sol!(
        #[sol(all_derives)]
        #[sol(extra_derives(serde::Serialize, serde::Deserialize))]
        #[sol(extra_methods)]
        #[sol(abi)]
        #[sol(rpc)]
        Pool,
        "abi/uniswap_v_3/pool.json",
    );

For pool events enum it can't derive std traits. The uniswap v3 pool has two different swap event, which could be the cause.

[satyakwok]

Dug into this — the trigger is the two overloaded Swap events, not the field types. Minimal repro (no JSON needed):

sol! {
    #[sol(all_derives)]
    contract Pool {
        event Swap(address indexed s, uint256 a);
        event Swap(address indexed s, uint256 a, uint256 b);
    }
}
// Pool::PoolEvents then lacks Debug / Eq / Hash (PartialEq + Clone are still there).

A single (non-overloaded) event with the same field types derives fine, so it is specifically the overload.

Root cause (in sol-macro-expander):

• For the events enum, ToExpand::Events::to_data sets variants to the overloaded names (overloaded_name, e.g. Swap_0 / Swap_1) and types: None, so generate_enum uses those overloaded idents as the variant types for type_derives.
• type_derives (all_derives branch) only adds [Debug, PartialEq, Eq, Hash] when can_derive_builtin_traits returns true for every variant type.
• can_derive_builtin_traits (ty.rs) does Type::Custom(name) => match self.try_item(name), but the overloaded idents (Swap_0) are not registered in all_items (the events live there under the original name Swap), so try_item returns None and it falls through to _ => false.
• derive_others becomes false, so Debug/Eq/Hash are dropped. Clone/PartialEq survive because they are added by the events path separately, which is why the failure looks partial.

Same mechanism affects overloaded errors, and it is the same reason the *Calls enum does not pick up all_derives builtin traits (related to alloy-rs/alloy#3857 / alloy-rs/alloy#1124).

Possible fix directions:

1. Make can_derive_builtin_traits resolve the overloaded variant name back to the underlying item, or
2. Compute the enum's derivability from the underlying event/error/function param types (which are available in to_data) instead of re-resolving the synthetic variant idents.

Happy to put up a PR — do you have a preference between (1) and (2)? I’d lean (2) since it fixes calls/errors/events uniformly without special-casing name resolution.

[Jinwei1987]

If alloy-rs/alloy#2 unify the derive stuff into single place, then alloy-rs/alloy#2 should be the right thing to do. I do notice to use other attribute to explicitly derive builtin traits don’t have such a issue.

…

On Thu, 11 Jun 2026 at 19:04, satya kwok ***@***.***> wrote: *satyakwok* left a comment (alloy-rs/core#1133) <#1133 (comment)> Dug into this — the trigger is the *two overloaded Swap events*, not the field types. Minimal repro (no JSON needed): sol! { #[sol(all_derives)] contract Pool { event Swap(address indexed s, uint256 a); event Swap(address indexed s, uint256 a, uint256 b); }}// Pool::PoolEvents then lacks Debug / Eq / Hash (PartialEq + Clone are still there). A single (non-overloaded) event with the same field types derives fine, so it is specifically the overload. Root cause (in sol-macro-expander): - For the events enum, ToExpand::Events::to_data sets variants to the *overloaded* names (overloaded_name, e.g. Swap_0 / Swap_1) and types: None, so generate_enum uses those overloaded idents as the variant types for type_derives. - type_derives (all_derives branch) only adds [Debug, PartialEq, Eq, Hash] when can_derive_builtin_traits returns true for *every* variant type. - can_derive_builtin_traits (ty.rs) does Type::Custom(name) => match self.try_item(name), but the overloaded idents (Swap_0) are not registered in all_items (the events live there under the original name Swap), so try_item returns None and it falls through to _ => false. - derive_others becomes false, so Debug/Eq/Hash are dropped. Clone/ PartialEq survive because they are added by the events path separately, which is why the failure looks partial. Same mechanism affects *overloaded errors*, and it is the same reason the *Calls enum does not pick up all_derives builtin traits (related to alloy-rs/alloy#3857 <alloy-rs/alloy#3857> / alloy-rs/alloy#1124 <alloy-rs/alloy#1124>). Possible fix directions: 1. Make can_derive_builtin_traits resolve the overloaded variant name back to the underlying item, or 2. Compute the enum's derivability from the underlying event/error/function param types (which are available in to_data) instead of re-resolving the synthetic variant idents. Happy to put up a PR — do you have a preference between (1) and (2)? I’d lean (2) since it fixes calls/errors/events uniformly without special-casing name resolution. — Reply to this email directly, view it on GitHub <#1133?email_source=notifications&email_token=AGWPBVFFG2LDFWF7G22FAWT47KG3BA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTINRXHE4DMMRSG442M4TFMFZW63VGMF2XI2DPOKSWK5TFNZ2KYZTPN52GK4S7MNWGSY3L#issuecomment-4843277484>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AGWPBVAA2GSDU2DVAIOYA3T47KG3BAVCNFSNUABFKJSXA33TNF2G64TZHM3DEOJSGQZTCNJSHNEXG43VMU5TIMJXGUYDIMZTGE2KC5QC> . Triage notifications, keep track of coding agent tasks and review pull requests on the go with GitHub Mobile for iOS <https://github.com/notifications/mobile/ios/AGWPBVH4T3ZCRGJK6VZQKUL47KG3BA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTINRXHE4DMMRSG442M4TFMFZW63VGMF2XI2DPOKSWK5TFNZ2KUZTPN52GK4S7NFXXG> and Android <https://github.com/notifications/mobile/android/AGWPBVA7Z35HXNN37TK4ZXD47KG3BA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTINRXHE4DMMRSG442M4TFMFZW63VGMF2XI2DPOKSWK5TFNZ2K4ZTPN52GK4S7MFXGI4TPNFSA>. Download it today! You are receiving this because you authored the thread.Message ID: ***@***.***>

[satyakwok]

Agreed — that's what #1118 does: it computes the enum's derivability from the underlying param types in one place, instead of re-resolving the synthetic overloaded variant idents. It's approved, so this should be fixed once it lands.