Repository navigation
Conversation
Stored conversation history is durable, but the save after a turn runs in the calling app after the workflow returns, not inside the workflow. A crash between the two loses the turn from memory, and a retry stores the user message twice. Say so wherever the docs, javadoc and the memory module's POM description describe chat memory as durable. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What changed
The docs said more about chat-memory durability than the code delivers. This narrows the claim to what is stored. The durable workflow wraps only the terminal model call (
DurableAdvisor). The chat-memory advisor sits outside it and saves in the calling app after the workflow returns. So stored history is persisted in the Dapr state store, but a memory save cut off by a crash is not replayed by the workflow.ConcurrentConversationModificationExceptionjavadoc: says the memory save runs outside the workflow and describes both retry paths. With a reused, caller-supplied instance id, the retry attaches to the completed workflow and returns the recorded answer without re-running the LLM call. With no supplied id, the retry runs a fresh workflow. Either way, the memory save on retry happens in the app again. This matches the instance-id table in the README.Calls made
DurableMemoryConciergesystem prompt ("a durable memory of the conversation") alone. It is text the model receives at runtime, not a claim to users, and changing it would change behaviour.AgentRecordFactory,AgentMetadataSchema) alone. It makes no claim about memory durability.Testing
Docs and javadoc only. Maven could not run here: the sandbox blocks
~/.m2and has no network access to Maven Central. Instead I checked the edited javadoc lines by hand against the CheckstyleLineLengthlimit (120); the longest is 102. Compile, tests, SpotBugs and PMD were not run, so CI'smvn -B clean installshould confirm.🤖 Generated with Claude Code
Review passes
Spec check
Checked against the issue as of: issue text hash
6db953f78811, latest comment no comments.Checks re-run by the worker
🤖 Generated with Claude Code