Expand description
Channels for state communication in graphs.
Channels are the core mechanism for state management in LangGraph. Unlike typical graph systems where nodes pass data directly, in LangGraph nodes write to channels and read from channels. This enables powerful patterns like automatic state reduction and checkpoint/resume.
§Channel Types
- LastValue: Stores only the last written value
- Topic: Accumulates all written values as a sequence
- BinaryOperatorAggregate: Reduces multiple writes with a custom function
- EphemeralValue: Cleared after each superstep
Re-exports§
pub use last_value::LastValue;pub use topic::Topic;pub use binop::BinaryOperatorAggregate;pub use ephemeral::EphemeralValue;
Modules§
- binop
- Binary operator aggregate channel implementation.
- ephemeral
- Ephemeral value channel implementation.
- last_
value - LastValue channel implementation.
- topic
- Topic channel implementation.
Traits§
- Base
Channel - The base trait for all channels.
Functions§
- binop
- Helper to create a BinaryOperatorAggregate channel
- last_
value - Helper to create a LastValue channel
- topic
- Helper to create a Topic channel
Type Aliases§
- Channel
Box - A wrapper for type-erased channels