A Tinybird engineer who has operated ClickHouse clusters since version 18.4 nearly six years ago published a detailed retrospective on running the database at petabyte scale. The post walks through the standard shard-and-replica architecture ClickHouse recommends and highlights the operational pitfalls that only surface once a cluster is running in production. Tinybird notes it is an independent contributor to the open-source project and unaffiliated with ClickHouse, Inc.
tinybird.co
· 2026-09-07
Immigration and Customs Enforcement has published documents outlining plans to award a contract worth up to $5 million for building a national database of voter registration and voting history records. The contract, managed by Homeland Security Investigations, would require the winning company to pull publicly available voter files from all 50 states, including data on which federal elections people voted in. ICE says the contractor won't initially be asked to collect party affiliation data, but that could change if a contracting officer directs it in writing.
wired.com
· 2026-09-02
SlideOps is a new pair of Agent Skills for Claude Code and compatible coding assistants that generates presentation slides directly from a codebase, then tracks the source each slide came from. When the underlying code changes, it detects which slides no longer match reality and rebuilds only those, rather than the whole deck. It installs via a two-line Claude Code marketplace command or a git clone plus install script for Codex, Copilot CLI, and OpenCode.
github.com
· 2026-08-31
An engineer upgrading a MySQL database via a replica discovered that a table's auto-incrementing primary key was assigned in a different row order on the replica than on the source. Because six related tables had been migrated to reference these new IDs, one of them ended up pointing to completely wrong rows after the switchover, while the other five happened to remain correct.
blog.elis.cc
· 2026-08-29
Researchers from Meta AI and the University of Illinois Urbana-Champaign built EvoHarness-RL, a training framework that teaches a small 8-billion-parameter model to decide when to read, update, or consolidate information it gathers from its execution environment during long, multi-step tasks. Rather than following rigid, human-written scripts for tool use, the model learns to manage its own memory and error recovery, reportedly reaching performance comparable to Claude Opus 4.5 on complex agentic workflows.
venturebeat.com
· 2026-08-28
A developer has released a lightweight database designed specifically for AI agent memory, built on a typed schema with supersession rules instead of relying on semantic text retrieval. The system aims to run cold-first at the edge, avoiding the cost and complexity of maintaining hot vector indexes in RAM or SSDs for smaller, resource-constrained deployments. The project was shared as a Show HN post on Hacker News.
polign.com
· 2026-08-26
Following up on his earlier SELF executable format that stores a program as a SQLite database, the author released self-httpd, a proof-of-concept web server whose code, website, routes, and visitor logs all live in a single SQLite file. The program reads its own instructions and writes new state, such as button-press logs, back into that same file transactionally, eliminating the need for separate folders like /var, /tmp, or /home.
fzakaria.com
· 2026-08-26
A developer created a workaround that adds rollback functionality to Flyway's free Community Edition, which normally reserves database migration undo capability for paying Teams subscribers. The trick treats every rollback as a forward migration: instead of reversing an applied migration, the tool writes a new, higher-numbered migration containing the reverse SQL, then cleans up Flyway's schema history table so the recorded state matches the database and files on disk. The implementation relies entirely on standard Flyway commands rather than direct SQL execution against the database.
aconcan.io
· 2026-08-25
LatticeDB is a new single-file, embedded property-graph database that supports graph traversal, HNSW-based vector similarity search, and BM25 full-text search through one unified query language. It requires no server or configuration, targets single-machine relationship-heavy workloads, and shares its transaction log across graph writes and event streams for durability.
github.com
· 2026-08-25
GoKawiil