Database Technology — OffNet Newsroom

GitHub Trending (daily) githubrepos ⚠ unverified date/source

WrenAI merges Wren Engine, expands GenBI context layer for AI agents

WrenAI is an open-source Generative BI tool that uses a governed context layer to translate natural language into trusted SQL and visualizations. It now integrates the previously separate Wren Engine directly into its core repository, supporting over 20 data sources including BigQuery, Snowflake, and Redshift. The legacy Docker-based chat interface is archived as Wren GenBI Classic on the v1-final tag.

Unified repo merges Wren Engine into WrenAI core, simplifying deployment and maintenance.
Supports 20+ data sources including BigQuery, Snowflake, PostgreSQL, and Redshift.
Legacy chat-first Docker app preserved as Wren GenBI Classic on legacy/v1 branch.
Focuses on governed text-to-SQL for AI agents with trusted dashboard generation.

Read at source →

Planet PostgreSQL 2026-07-20 database

PostgreSQL extra_float_digits: Precision without the tradeoff since v12

Christophe Pettus explains how the extra_float_digits GUC evolved from a forced tradeoff between readability and precision to a setting that supports both. Prior to version 12, administrators had to choose between human-readable output and exact floating-point representation. Modern PostgreSQL versions handle this automatically, removing the need for manual tuning of this specific parameter.

PostgreSQL 12+ eliminates the historical choice between readable and exact float output
extra_float_digits no longer requires manual tuning for precision in modern versions
Review legacy configs if migrating older deployments to ensure no outdated workarounds remain
Verify application serialization logic relies on default behavior for consistency

Read at source →

AWS What's New 2026-07-16 awsdatabase ↺ since 07-17

AWS RDS previews PostgreSQL 19 Beta 2 with parallel autovacuum and native graph queries

Amazon RDS has opened the Database Preview Environment for PostgreSQL 19 Beta 2, allowing users to test pre-release features on a fully managed platform. Key additions include parallel autovacuum with configurable worker limits to prevent maintenance bottlenecks and a new REPACK CONCURRENTLY command for online storage reclamation without third-party tools. The release also introduces native SQL Property Graph Queries for relationship traversals and updates to logical replication synchronization.

Parallel autovacuum with configurable workers prevents maintenance from stalling large databases.
REPACK CONCURRENTLY enables online table rebuilding and storage reclamation without extensions.
Native SQL/PGQ support allows relationship traversals directly in standard SQL.
Logical replication synchronization features have been updated in this beta release.
Preview Environment provides a safe sandbox to evaluate RDS-specific behavior for PG19.

Read at source →

InfoQ 2026-07-15 generaldevops ↺ since 07-16

Postgres for Production AI Agents: Vector Indexing and Memory Strategies

Gwen Shapira outlines how PostgreSQL serves as a relational foundation for enterprise AI agents by leveraging JSONB and HNSW vector indexing. She highlights techniques like vector quantization to achieve four-times faster query performance while maintaining semantic context for LLMs. The presentation also covers practical approaches to managing agentic memory within mission-critical applications.

Use HNSW vector indexing in Postgres for high-recall semantic search capabilities.
Apply vector quantization to accelerate vector queries by approximately 4x.
Combine JSONB parsing with vector data to provide deterministic and semantic context to LLMs.
Implement structured strategies for managing agentic memory within the database layer.

Read at source →

Planet PostgreSQL 2026-07-17 database ↺ since 07-18

Postgres 19 Introduces Granular Manual Checkpoint Control

PostgreSQL 19 enhances the manual CHECKPOINT command, moving beyond the previous all-or-nothing immediate flush behavior. This update addresses the need for more controlled write-back operations, which is critical when background checkpointing is already straining storage resources. DBAs can now invoke targeted flushes with finer granularity to manage I/O loads effectively.

Postgres 19 expands manual CHECKPOINT capabilities beyond simple immediate flushing.
Helps mitigate storage overload when background checkpointing is under stress.
Enables DBAs to trigger targeted write-backs with better I/O control.
Improves predictability for maintenance windows requiring data persistence.

Read at source →

Planet PostgreSQL 2026-07-16 database ↺ since 07-17

Postgres 19 shifts default TOAST compression from pglz to LZ4

PostgreSQL 19 plans to replace the legacy pglz algorithm with LZ4 for automatic TOAST and heap compression of variable-length types. The database employs a unified compression framework where heap and TOAST compression is always active, while index compression remains opportunistic based on key size thresholds. This change updates the internals of table storage and indexing without altering the automatic nature of compression for large fields.

LZ4 replaces pglz as the default for TOAST and heap compression in Postgres 19.
Compression remains automatic for TEXT, VARCHAR, BYTEA, and JSONB types.
Index compression stays opportunistic, triggering only when keys exceed size thresholds.
All storage layers use a unified compression framework for consistency.

Read at source →