⚠️ No Changeset found

Latest commit: 34aebe9

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR

Greptile Overview

Greptile Summary

This PR fixes critical error handling issues in the eval runner that were causing tasks to idle and memory leaks.

Key improvements:

• Moved v3Input declaration to outer scope so it's accessible in all error paths
• Removed nested try/finally that was complicating cleanup logic
• Added comprehensive finally block that guarantees V3 instance closure and logger cleanup
• Close errors are now logged as warnings instead of throwing, preventing them from masking original task results
• Added clear() method to EvalLogger to free memory after logs are captured

The changes follow clean error handling patterns: logs are captured in return statements before the finally block clears them, and cleanup always happens regardless of success or failure.

Confidence Score: 5/5

• This PR is safe to merge with no risk
• The changes implement robust error handling and resource cleanup patterns. The variable scoping is correct, cleanup is guaranteed via finally block, and error handling properly prevents cleanup errors from masking original results. No edge cases or issues found.
• No files require special attention

Important Files Changed

File Analysis

Sequence Diagram

sequenceDiagram
    participant Runner as Eval Runner
    participant Logger as EvalLogger
    participant Task as Task Module
    participant V3 as V3 Instance

    Runner->>Logger: new EvalLogger()
    Note over Runner: Declare v3Input at outer scope
    
    alt Task Execution Success
        Runner->>Task: import task module
        Runner->>V3: initV3()
        V3-->>Runner: v3Input
        Runner->>Task: taskFunction(v3Input)
        Task->>Logger: log events
        Task-->>Runner: result { _success: true }
        Runner->>Runner: console.log success
        Runner-->>Runner: return result
    else Task Execution Error
        Runner->>Task: import or execute task
        Task--xRunner: throws error
        Runner->>Logger: logger.error()
        Runner-->>Runner: return { _success: false, error, logs }
    end
    
    Note over Runner,V3: Finally block always runs
    Runner->>V3: v3Input?.v3.close()
    alt Close Success
        V3-->>Runner: closed
    else Close Error
        V3--xRunner: closeError
        Runner->>Runner: console.error (warning only)
    end
    Runner->>Logger: logger.clear()
    Logger->>Logger: logs = [], stagehand = undefined