fix: atomic incremental snapshots and reliable CLI output (1.17.1) - #42
Merged
Merged
Conversation
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Incremental CLI indexing previously modified the active vector table before committing metadata, so terminating the process could leave inconsistent counts. Stage changed indexes and atomically activate their metadata after validation; no-change runs retain the fast path and unchanged embeddings are reused.
Also apply duplicate
--limitto every CLI JSON projection, report output truncation separately, default search to one result per file, and preserve domain words in cluster labels. Release core/CLI/MCP 1.17.1 with updated documentation.Validation: typecheck; full 238-test suite plus the added no-op regression (14 indexer tests); real Windows/Ollama process termination during persistence preserved the old snapshot and recovered; 32 real search calls improved default expected-file top-five coverage from 13/16 to 15/16; duplicate limits in minimal/diagnostic/full, context budget, graph, labels and strict index health passed. Individual watcher mutations remain outside the staging change; broad search is still not exhaustive.