Struct Graph 

pub struct Graph {
    pub nodes: Vec<Node>,
    pub edges: Vec<Edge>,
    pub training_strategy: Option<TrainingStrategy>,
}

A directed graph of computational nodes.

Fields

nodes: Vec<Node>

The nodes, in insertion order (execution order comes from the edges).

edges: Vec<Edge>

The directed edges connecting them.

training_strategy: Option<TrainingStrategy>

Training strategy for distributed execution. Inherited by subgraphs unless overridden.

Implementations

impl Graph

pub fn new() -> Self

Create an empty graph with no training strategy set.

pub fn with_strategy(self, strategy: TrainingStrategy) -> Self

Set the training strategy for this graph.

pub fn set_strategy(&mut self, strategy: TrainingStrategy)

Set the training strategy (mutable).

pub fn effective_strategy_is_distributed(&self) -> bool

Whether the effective strategy needs more than this process.

Local does not, and neither does an absent one — which is the same thing. Everything else asks for workers, so a caller can use this to decide whether to take the distributed path at all rather than matching on the enum in three places.

pub fn effective_strategy(&self) -> &TrainingStrategy

Get the effective training strategy (defaults to Local).

pub fn add_node(&mut self, node: Node)

Add a node. Duplicate ids are not checked here; Self::validate rejects them at compile time.

pub fn add_filter(&mut self, filter_name: impl Into<String>) -> &str

Add a filter node using the filter name as the node id. If a node with that name already exists, appends a suffix.

pub fn add_edge(&mut self, edge: Edge)

Add an edge. Endpoints are not checked here; Self::validate rejects edges to unknown nodes at compile time.

pub fn connect(&mut self, source: impl Into<String>, target: impl Into<String>)

Connect two nodes with a data edge (auto-generates edge id).

pub fn node(&self, id: &str) -> Option<&Node>

Get a node by its ID.

pub fn node_ids(&self) -> Vec<&str>

Get all node IDs.

pub fn predecessors(&self, node_id: &str) -> Vec<&str>

Get predecessors of a node (nodes with edges pointing to it).

pub fn successors(&self, node_id: &str) -> Vec<&str>

Get successors of a node (nodes it points to).

pub fn roots(&self) -> Vec<&str>

Find root nodes (no incoming edges).

pub fn leaves(&self) -> Vec<&str>

Find leaf nodes (no outgoing edges).

pub fn topological_sort(&self) -> Result<Vec<&str>>

Topological sort using Kahn’s algorithm. Returns Err if the graph contains a cycle.

pub fn validate(&self) -> Result<()>

Validate the graph structure (recursively validates sub-graphs).

pub fn contains_steps(&self) -> bool

Does this graph — or any sub-graph nested inside it — contain a step?

A step calls models and tools, so a graph that contains one is not a deterministic function of its input. crate::effect::Effect::is_pure asks this before memoizing a graph effect by content.

impl Graph

pub fn to_mermaid(&self) -> String

Render as a Mermaid diagram.

graph LR
    scaler[scaler]
    model[model]
    scaler --> model

pub fn to_mermaid_with(&self, overlay: &GraphOverlay) -> String

Render as a Mermaid diagram with per-node execution annotations.

Each annotated node gets a second label line (duration, cache tier, health flags — see crate::viz::NodeOverlay::sublabel_text) and a status classDef for coloring. An empty overlay produces exactly Graph::to_mermaid’s output.

pub fn to_graphviz(&self) -> String

Render as Graphviz DOT format.

pub fn to_graphviz_with(&self, overlay: &GraphOverlay) -> String

Render as Graphviz DOT with per-node execution annotations: a second label line plus fill/border status colors. An empty overlay produces exactly Graph::to_graphviz’s output.

pub fn to_text(&self) -> String

Render as an ASCII text tree for terminal display.

impl Graph

pub fn to_svg(&self) -> String

Render as a self-contained SVG diagram.

pub fn to_svg_with(&self, overlay: &GraphOverlay) -> String

Render as a self-contained SVG diagram with per-node execution annotations (status colors + a duration/cache/flags sublabel), same overlay semantics as Graph::to_mermaid_with.

Trait Implementations

impl Clone for Graph

fn clone(&self) -> Graph

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for Graph

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for Graph

fn default() -> Self

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for Graph

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Display for Graph

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Serialize for Graph

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.