Elior Nataf Lackritz 8c37ffb062 fix(langgraph): never store exit-mode delta writes under the null task id ...

exit_delta_task_id puts the superstep in the first UUID group. For a
Command's writes in a run whose first superstep is 0, such as one that
starts from an update_state(..., as_node="__input__") checkpoint, that
gave NULL_TASK_ID itself, so get_state and later runs took the writes for
the checkpoint's own pending writes. The DeltaChannel then read the update
on that checkpoint and a replay from it applied it again, while a plain
channel did neither.

2026-10-07 11:38:51 -04:00

..

checkpoint

fix(langgraph): replay a resumed exit-mode run's delta writes in live order (#9114)

2026-10-06 22:05:55 +00:00

checkpoint-conformance

fix: order delta channel replay by task path (#8544)

2026-10-06 17:55:04 -04:00

checkpoint-postgres

fix: order delta channel replay by task path (#8544)

2026-10-06 17:55:04 -04:00

checkpoint-sqlite

fix: order delta channel replay by task path (#8544)

2026-10-06 17:55:04 -04:00

cli

release(cli): 0.4.33 (#9227)

2026-10-07 09:19:34 -04:00

langgraph

fix(langgraph): never store exit-mode delta writes under the null task id

2026-10-07 11:38:51 -04:00

prebuilt

fix: order delta channel replay by task path (#8544)

2026-10-06 17:55:04 -04:00

sdk-js

docs: standardize package README.md structure (#8064)

2026-06-12 19:32:10 -04:00

sdk-py

fix: order delta channel replay by task path (#8544)

2026-10-06 17:55:04 -04:00