Struct LoopNode 

pub struct LoopNode {
    pub name: String,
    pub body: SubGraph,
    pub continue_condition: Arc<dyn Fn(&State) -> bool + Send + Sync>,
    pub max_iterations: usize,
}

循环容器 — 可选的高级语法糖。

推荐使用 edge_if 实现简单回跳。 LoopNode 适用于需要独立迭代计数 和独立熔断保护的封装场景（例如并行子任务中的局部循环）。

// 推荐：直接用有环图 + edge_if（更直观）
GraphBuilder::new("retry")
    .edge_if("check", "agent", |s| !s.satisfied)  // 回跳
    .edge("check", "output")                       // 通过

// LoopNode：需要独立 max_iterations 时使用
LoopNode::new("loop", SubGraph { ... }, |s| !s.satisfied, max_iterations: 5)

Fields

name: String

body: SubGraph

continue_condition: Arc<dyn Fn(&State) -> bool + Send + Sync>

max_iterations: usize

Implementations

impl LoopNode

pub fn new( name: impl Into<String>, body: SubGraph, continue_condition: impl Fn(&State) -> bool + Send + Sync + 'static, max_iterations: usize, ) -> Self

Trait Implementations

impl GraphNode for LoopNode

fn execute<'life0, 'life1, 'async_trait>( &'life0 self, state: &'life1 mut State, ) -> Pin<Box<dyn Future<Output = Result<NextStep, GraphError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

执行节点逻辑（阻塞模式）。

fn execute_stream<'life0, 'life1, 'life2, 'async_trait>( &'life0 self, state: &'life1 mut State, _sink: &'life2 Sender<GraphEvent>, span_id: SpanId, ) -> Pin<Box<dyn Future<Output = Result<StreamNodeResult, GraphError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait,

执行节点逻辑（流式模式），将内部事件转发到 channel。