Skip to main content

EvalExecutor

vtcode_eval::executor

Trait EvalExecutor 

Source 
pub trait EvalExecutor: Send + Sync {
    // Required method
    fn execute_task<'life0, 'life1, 'async_trait>(
        &'life0 self,
        task: &'life1 EvalTask,
    ) -> Pin<Box<dyn Future<Output = Result<EvalRunResult>> + Send + 'async_trait>>
       where Self: 'async_trait,
             'life0: 'async_trait,
             'life1: 'async_trait;
}
Expand description

Executes a single eval task attempt and returns the verified outcome.

Implementations own the full “execute this task” semantics — running the agent and applying any environment verification (probes). Keeping this behind a trait decouples the orchestration in run_suite from the concrete runner, which makes the harness independently testable.

Required Methods§

Source

fn execute_task<'life0, 'life1, 'async_trait>( &'life0 self, task: &'life1 EvalTask, ) -> Pin<Box<dyn Future<Output = Result<EvalRunResult>> + Send + 'async_trait>>
where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

Run one attempt of task and return the outcome.

Dyn Compatibility§

This trait is dyn compatible.

In older versions of Rust, dyn compatibility was called "object safety".

Implementors§