#244 · Measure search query plans on realistic synthetic data
Last checked with GitLab: . Status comes from GitLab using a 2-minute cache. Merge events clear affected cached statuses. An already-open page does not update itself. Refresh page.
An MR completes this bounty when it is merged into main with Closes #244 and GitLab closes the issue. A mention or an updated timestamp is not completion. Awarded points appear in Activity.
Description
Search optimization starts from measured bottlenecks and recall behavior.
Who might enjoy this
- Skills: SQL, Data.
- Experience: intermediate. Work style: analytical.
- Readiness: Ready. Size: M (existing scoreboard tier; standard merge award stacks separately).
Current behavior and entry points
Source-reviewed against main cd4cc95b on 2026-10-18. Re-read current main and open MRs before implementation; preserve working features and narrow the change to the gap described here.
Relevant Files
apps/search-service/src/sources/d1docs/search-scale-roadmap.md
Dependencies and decisions
None; rebase on current main and check for a linked MR before starting.
Acceptance criteria
- [ ] Seed a repeatable local dataset at current and documented next-scale sizes and record query plans/timings.
- [ ] Verify known matches and no-match controls before reporting performance.
- [ ] Submit a measured report and at most one justified index change; no production benchmark, chat backfill or premature vector migration.
- [ ] Record named baseline failures and real verification commands. For a behavior fix, demonstrate the regression check fails on old behavior and passes after the change. Documentation/device work includes reproducible evidence appropriate to its scope.
- [ ] Run the affected package's existing typecheck, test and lint scripts where available; compare named failures with main. Use a real router/local database/protocol path when that is the boundary being changed.
Hints for your coding agent
Read AGENTS.md, the applicable .cursor/rules/ file, app CLAUDE.md, and docs/contributing/agentic-bounty-review.md. Reuse the existing shared helpers, identity sources and app styling. Explain the member outcome and the exact change before implementing. Work in an isolated branch; one thin, independently reviewable MR. The listed paths are entry points, not permission to rewrite their whole area.
Out of scope
Unrelated refactors, a new framework/provider, bulk dependency updates, production access changes and unrequested notifications. Never commit credentials, private member data, raw production logs or unresolved security findings. Member content stays local-first. A live/device check needs an operator-approved test environment; this bounty is not blanket authority to change production.
Review and award
The accepting reviewer is the app maintainer using the repository review process. Do not self-approve or claim completion from HTTP 200, green CI, a mock alone or a process exit alone. Record updated, pre-release verified, released and live verified accurately. A parent tracker and child story must not award twice for the same work. Open a scoped MR closing this issue; only its actual issue ID belongs after the closing keyword. Link your GitLab identity on Scoreboard so the existing award flow can credit you.
<!-- community-bounty-catalog:2026-10:query-plan -->
Catalog index: #191.
Relevant files
-
apps/search-service/src/sources/d1from issue -
docs/search-scale-roadmap.mdfrom issue
Hints (1)
-
Read
from issueAGENTS.md, the applicable.cursor/rules/file, appCLAUDE.md, anddocs/contributing/agentic-bounty-review.md. Reuse the existing shared helpers, identity sources and app styling. Explain the member outcome and the exact change before implementing. Work in an isolated branch; one thin, independently reviewable MR. The listed paths are entry points, not permission to rewrite their whole area.
Activity
- +100 Hugo Zarate earned M bounty 6h ago
Comments
Sign in to join the discussion.
No comments yet. Be the first to weigh in.