mirror of
https://github.com/langchain-ai/langgraph.git
synced 2026-10-20 15:01:44 +02:00
The async loop skipped caching an interrupted or failed task, but the sync loop didn't, so a cached error made the next run with the same input skip the node without raising. Both loops now share the check, which looks at every write, since a cancelled task's error comes after any writes it already made. Also drops `arun_with_retry`'s `match_cached_writes`, unused since #4691 moved cache matching into the loop.