Repository navigation
test: cover StructuredDiscussion.flatten context truncation and cache breakpoints - #34
Open
devin-ai-integration[bot] wants to merge 1 commit into
Open
devin-ai-integration[bot] wants to merge 1 commit into
devin-ai-integration[bot] wants to merge 1 commit into
Conversation
…ints (bug risk in untested index-sensitive logic)
Author
🤖 Devin AI EngineerI'll be helping with this pull request! Here's what you should know: ✅ I will automatically:
Note: I can only respond to comments from users who have write access to this repository. ⚙️ Control Options:
|
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Closes a code-scan coverage gap:
StructuredDiscussion.flattenbuilds the exact message list sent to the model on everyToolUseAgentstep, and had no test anywhere. A regression there silently changes model context (dropped observations) or prompt caching (cost/correctness), with no failing test.New file
tests/agents/test_tool_use_agent.py, pure objects, no LLM and no network. It pins both halves of the index-sensitive logic:keep_last_n_obs— non-tail groups collapse togroup.summary, the lastngroups stay verbatim, and a non-tail group without a summary falls back to its messages:_cache_breakpointflags (including a stale one pre-set on an earlier message) are cleared and exactly one breakpoint is set at the tail boundary; whenkeep_last_n_obsexceeds the group count the boundary index is negative and no breakpoint is set at all.Tests use the real
AnthropicAPIMessageBuilder(the only builder that actually implementsmark_all_previous_msg_for_cachingvia_cache_breakpoint) rather than stubs, so the assertions track real message objects.Note for reviewers:
flattenmarksmessages[i]whereiis the group index, which coincides with the first message of the boundary group only when every collapsed group contributes exactly one message. The tests document current behavior; changing that indexing is out of scope here.Devin-Org: engineering
Link to Devin session: https://app.devin.ai/sessions/5a670129ef3840b1ad2fd78b2267b90c
Open in Devin Desktop: https://app.devin.ai/desktop/session/5a670129ef3840b1ad2fd78b2267b90c?variant=devin