Releases: langchain-ai/langgraph
Releases · langchain-ai/langgraph
0.2.67
Changes since 0.2.66
- langgraph: release 0.2.67 (#3182)
- langgraph: update docstrings/api ref for functional api (#3176)
- langgraph: handle node return annotations with unions (#3170)
- langgraph: add support for BaseModel updates to Command (#2747)
- langgraph: remove print (#3167)
- Make scratchpad counters thread-safe (#3158)
- Make test less flaky
- Add entrypoint.final to decouple return value from save value (#3135)
- langgraph: allow async state modifier in create_react_agent (#3161)
- Undo
- Fix
- Update tests
- Make scratchpad counters thread-safe
- Lint
- Fix
- Fix
- Fix
- add test case
- Lint
- Format
- Remove write to RETURN channel for entrypoint func
- Disable recursing on runnables for task/entrypoint decorated funcs
- Spell check
0.2.66
Changes since 0.2.65
- langgraph: release 0.2.66 (#3128)
- docs: update readme / docs intro page (#3082)
- Enable async tests that were being skipped (#3109)
- Fix timing issue where a sync task would finish before the other one was registered in futures dict (#3110)
- Re-enable support for running sync tasks from async entrypoints (#3108)
- Fix tracing of args for @task decorated functions (#3107)
- Enable async tests that were being skipped
- Fix timing issue where a sync task would finish before the other one was registered in futures dict
- Re-enable support for running sync tasks from async entrypoints
- Fix tracing of args for @task decorated functions
0.2.65
cli==0.1.68
Changes since cli==0.1.67
- feat(cli): add detection for bun.lockb
- fix(cli): warn users to use the JS cli for JS graphs
0.2.64
Changes since 0.2.63
- langgraph: release 0.2.64 (#3090)
- tests: add tests for calling multiple subgraphs in a parent node (#3070)
- functional api: first pass at api reference (#3083)
- docs: add high level example into the readme, make examples collapsible (#3052)
- Make config schema configurable for imperative api
- Implement input/output schemas for imperative api
- tests: add a test for interrupt() w/ functional API
0.2.63
Changes since 0.2.62
- langgraph: release 0.2.63 (#3064)
- tests: Add test for catching bunching in streaming (#3057)
- Add support for multiple subgraphs called in a single node (#3056)
- Fix kafka lib
- Add checkpointer=True mode for subgraphs that want to keep state between turns (#3055)
- Add support for multiple subgraphs called in a single node
- Add checkpointer=True mode for subgraphs that want to keep state betweenn turns
- Fix unexpected re-use of null resume value by subgraphs (#3054)
- Fix ignored goto when a mixed list of command and state updates is returned from a node (#3038)
- Fix unexpected re-use of null resume value by subgraphs
- feat(langgraph): Add interrupt schema to library (#2947)
- Fix Send order after interrupt/resume (#3037)
- tests: add test for multiple interrupts and tasks (#2941)
- Fix Send order after interrupt/resume
- Fix multiple interrupt/task test
- Remove duckdb checkpointer and store
- Fix flasy return from task
- Fix tracing hierarchy for imperative api (#3036)
- Remove duckdb checkpointer and store
- langgraph: fix non-empty value check in ensure_config (#3039)
- Remove send v2 (#3033)
- langgraph: allow model names as strings in create_react_agent (#3031)
- docs: document inject kwarg (#3026)
- Fix tracing hierarchy for imperative api
- Remove send v2
- docs: internal doc nits (#2946)
- langgraph: fix ismethod check in add_node (#3032)
- feat: Add interrupt schema to library
checkpointpostgres==2.0.13
checkpointsqlite==2.0.3
Changes since checkpointsqlite==2.0.2
- checkpoint-postgres/sqlite 2.0.12/2.0.3
- Update sqlite signature
checkpointpostgres==2.0.12
Changes since checkpointpostgres==2.0.11
- checkpoint-postgres/sqlite 2.0.12/2.0.3
- Add optional task_path arg for put_writes()
- checkpoint-duckdb: handle calling .list on async checkpointer (#3022)
checkpoint==2.0.10
Changes since checkpoint==2.0.9
- checkpoint 2.0.10
- Add optional task_path arg for put_writes()