Skip to content

Add optional local search index cache#60

Closed
Xuanwo wants to merge 1 commit into
mainfrom
xuanwo/search-index-cache-architecture
Closed

Add optional local search index cache#60
Xuanwo wants to merge 1 commit into
mainfrom
xuanwo/search-index-cache-architecture

Conversation

@Xuanwo

@Xuanwo Xuanwo commented Mar 24, 2026

Copy link
Copy Markdown
Owner

This PR turns xurl's query acceleration into a layered local cache with provider discovery manifests, thread materialization reuse, and an internal FTS-backed search index maintained by the hidden xurl -X index worker. It keeps correctness on the direct-scan path while letting repeated searches and rebuilds stay fully local and disposable.

The change also adds a repository benchmark at xurl-core/benches/search_index.rs so we can track regressions in cold query, warm query, incremental refresh, full build, and FTS rebuild. On the current synthetic codex benchmark (2500 threads, 3 samples), warm query is about 9.6ms, incremental refresh about 451ms, full build about 647ms, and FTS rebuild about 556ms.

@Xuanwo

Xuanwo commented Mar 24, 2026

Copy link
Copy Markdown
Owner Author

Nah, doesn't work great.

@Xuanwo Xuanwo closed this Mar 24, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant