Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

1.0.0 - 2026-03-21

Added

• Three-tier prepared statement cache metrics: request hits, connection hits, and true PG misses
• Process memory (RSS/VSZ) and Tokio runtime gauges (alive tasks, workers, global queue depth)
• DataLoader for batching forward relation lookups
• Per-request DB client (RequestClient) sharing a single pooled connection across all resolvers
• Grafana dashboard with dynamic pool utilization gauge
• Prometheus scrape config and hey_blast.sh parallel load testing
• Full documentation site (fumadocs)
• --metrics now defaults to true

Fixed

• Statement cache metrics: previously all deadpool lookups were counted as misses (96% false miss rate), now correctly shows 99.99% hit rate
• graphql_errors_total counter initialized at startup to prevent NaN in Prometheus rate() expressions
• DB pool max_size gauge seeded at startup

Removed

• Schema reload metrics (schema_reloads_total, schema_reload_duration_seconds, schema_reload_failures_total)

0.9.0 - 2026-03-19

Fixed

• Comprehensive PostgreSQL type serialization -- PostGraphile-compatible GraphQL types for UUID, Time, Interval, Point, BitString, and InternetAddress; wire list filter types for array columns (#9)

0.8.0 - 2026-03-19

Fixed

• NUMERIC columns returning null -- switched to rust_decimal for NUMERIC column deserialization; fixed BigFloat filter params (#8)

0.7.0 - 2026-03-19

Fixed

• Synthetic foreign keys for historical tables -- create synthetic FKs from table comments so relations resolve correctly on _block_range tables (#7)

0.6.1 - 2026-03-19

No user-facing changes (tag co-located with v0.6.0).

0.6.0 - 2026-03-19

Added

• Fulltext search support via @fullText directive -- sanitize_tsquery input sanitization, ts_rank ordering (#6)

Fixed

• Create synthetic FKs from table comments for historical tables
• Read smart tags from table comments (historical mode)

0.5.0 - 2026-03-19

Added

• PostGraphile @derivedFrom smart tag support -- backward relation field names resolved via column and table comment smart tags (#5)

0.4.0 - 2026-03-19

Fixed

• GraphiQL playground and route conflict on / -- playground now serves correctly alongside the GraphQL endpoint (#4)

0.3.0 - 2026-03-19

Changed

• Serve GraphQL on / instead of /graphql to match PostGraphile routing (#3)

Fixed

• Use git tag for GitHub Release name instead of Cargo.toml version

0.2.0 - 2026-03-19

Added

• Order by related entity scalar fields -- pg-order-by-related parity with correlated subquery ORDER BY enum values (#2)

0.1.0 - 2026-03-19

Added

• Initial Rust implementation of omnihedron (#1)
• Full dynamic GraphQL schema generation from PostgreSQL introspection
• Connection queries with cursor-based and offset pagination
• Filter input types with per-column operators (equalTo, in, like, isNull, logical and/or/not)
• Forward and backward relation fields (FK resolution)
• Relation filters (some/none/every, Exists boolean)
• Aggregation queries (sum, count, min, max, average, stddev, variance)
• Historical table support (_block_range detection, blockHeight argument)
• _metadata and _metadatas multi-chain queries
• node(nodeId) and {entity}ByNodeId root queries (Node interface)
• Batch query support (POST with JSON array)
• GraphQL subscriptions via WebSocket
• Schema hot reload (LISTEN/NOTIFY with atomic Arc<RwLock<Schema>> swap)
• Selective SELECT -- only columns referenced in the query are fetched
• Count-only fast-path -- queries without nodes/edges skip row fetch
• Window function COUNT(*) OVER() for single-roundtrip total count
• Selective aggregate computation (only requested aggregates in SQL)
• Query validation: complexity, depth, alias, and batch limits
• Cache-Control headers (public, max-age=5)
• HTTP request tracing with UUID request IDs
• DB_HOST_READ read replica support
• --indexer metadata HTTP fallback
• /health endpoint
• Docker image published to polytopelabs/omnihedron
• 62 integration tests across 8 test files
• CI workflows for formatting, integration tests, and Docker publishing