jordan.washburn
fe711870be
feat: M8 phases 7-10 — gRPC transport, cluster server, scatter-gather, multi-node UAT
...
Delivers the distributed fabric's network layer and HTTP cluster surface:
**m8p7: tidal-net crate (gRPC transport)**
- GrpcTransport implementing Transport trait via tonic 0.12
- Per-peer circuit breaker (Closed/Open/HalfOpen), mutual TLS via rustls
- Boxed error types (clippy-clean), graceful mutex recovery, debug_assert
against calling block_on from tokio context
- Proto: WalShipping service (ShipSegment, StreamSegments stub, Heartbeat)
- 19 tests: contract, mTLS, reconnection, multi-node UAT, benchmarks
**m8p8: cluster subcommand + HTTP routes**
- ClusterState wrapping SimulatedCluster with region name mapping
- Routes: /health, /cluster/status, /cluster/promote, /partition, /heal
- Data routes: /items, /embeddings, /signals, /feed, /search (region-aware)
- Ranking profiles wired through ClusterConfig to all cluster nodes
- Topology YAML config, docker/cluster/Dockerfile (ENTRYPOINT+CMD, non-root)
**m8p9: scatter-gather query routing**
- Entity-sharded writes via Knuth multiplicative hash
- Scatter-gather RETRIEVE and SEARCH with deadline propagation (50ms-5ms)
- Partial failure: degraded=true with unavailable_shards metadata
- 6 tests: distribution, determinism, multi-shard retrieve, degraded
partial results, deadline propagation, scatter-gather search
**m8p10: gRPC transport integration tests**
- 8 tests over real gRPC: replication convergence, idempotent replay,
mixed signals, 3-node fan-out, partition/heal, degraded follower, perf
- Documented as tier-2 (in-process+gRPC); tier-3 multi-process pending
2026-04-11 13:51:08 -06:00
jordan
006d3d058a
fix: populate signal_snapshot in ranking results and fix quickstart demo
...
- Score and boost contributions now captured per candidate in signal_snapshot:
score_by_sort() returns (f64, Vec<(String, f64)>) with named signal values
(e.g. view_velocity, share_velocity, view, like). Boost contributions are
appended as `{signal}_boost`. compute_raw_score() threads the snapshot
through; score_inner() and score_personalized() write it onto ScoredCandidate
instead of always initialising to vec![].
- Quickstart switched from `trending` to `hot` profile. The trending profile
scores by 24h velocity, which requires signals to arrive over real elapsed
time to populate hour-level bucket aggregates — impossible in a self-contained
demo. The hot profile (AllTime view count + age decay) works with any timestamp
and produces clearly differentiated scores. Signals now use different counts
per item to drive meaningful ranking output.
2026-03-10 17:06:21 -06:00
jordan
192c473f55
feat: complete Milestone 5 — full-text search, RRF fusion, and creator search
...
- M5p1: BM25 text indexing via Tantivy with background syncer (0.26ms @ 10K docs)
- M5p2: RRF fusion layer combining BM25 + ANN scores (46µs @ 1K candidates)
- M5p3: unified Search query API (8-stage pipeline, BM25 + vector + ranking)
- M5p4: creator text + vector indexing and creator search executor (< 20ms @ 200 creators)
- Refactor db/mod.rs into focused sub-modules (creators, items, sessions, signals, etc.)
- Decompose monolithic files into directory modules (query/executor, ranking/diversity, etc.)
- Split brute.rs → brute/mod.rs + brute/tests.rs; extract search executor helpers
- Add benches: fusion, search, session, text_index
- Add M5 UAT test suites (m5_uat, m5_search, m5p4_creator_search, text_index)
- Update blog posts, roadmap, content strategy, and M5 planning docs
- Add tmp/ and .claude/worktrees/ to .gitignore
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-02-21 23:53:16 -07:00
