fix: preserve live context during memory compression - #1048
Open
vaibhavmashal wants to merge 1 commit into
Open
Conversation
|
|
Contributor
|
✅ OpenCodeReview: Review complete: 0 finding(s) across 1 selected item(s). |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Problem
Three-zone compression (
internal/llmloop/compression.go) discards live context it is designed to preserve, in three related ways:This leads to unintended truncation and summarization of recent context.
Solution
This PR applies the following fixes:
compressEnd == frozenEnd, rather thancompressEnd == len(messages).messages[:frozenEnd]in addition toprevSummaryTokenEstimatewhen computing the active zone size.runCompressionreturn the conversation unchanged when no template is configured instead of slicing it down to the first 2 messages.Changes Made
internal/llmloop/compression.go:partitionMessagesto compute and pass the frozen zone token count tocomputeActiveZoneSize.partitionMessagesto setcompressEnd = result.frozenEndinstead oflen(messages)when everything fits in the active zone.runCompressionto returnmsgsinstead ofmsgs[:min(len(msgs), 2)]when the compression template is empty.internal/llmloop/compression_zone_repro_test.go: Added reproduction tests to ensure these edge cases are covered.Testing
TestRepro_EverythingFitsSummarizesLiveTail,TestRepro_FrozenZoneNotReservedFromBudget, andTestRepro_MissingTemplateTruncatesConversation.Fixes #838