In this article, I would like to share my experience participating in the Agentic Legal RAG Challenge 2026 hackathon. Our team is called "Sparks of intelligence".
Original article in Russian: https://habr.com/ru/articles/1014520/
Build a self-hosted VPN for full control, better privacy, and reliable access, even in regions with censorship and DPI-based blocking
If you are using a LoRa module for point-to-point (P2P) communication or building your own private protocol, understanding and correctly configuring each parameter is the key to successful communication.
After reading this article, you will learn:
• How Should the LoRa Over-the-Air (PHY) Modulation and Packet Format Parameters Be Configured, and Why?
• The meaning of RF front-end parameters and how to tune them
• Chip-level control parameters: operating modes, clocks, power supply, calibration, and more
• A quick-reference "must-match" table and a list of common issues
Note: This article covers only LoRa (P2P/Private Protocol) parameters. If you need information on LoRaWAN protocol-layer parameters (e.g., ADR, DevEUI, join procedures), please refer to "LoRaWAN Protocol Layer Parameters".
Before starting configuration, verify that your device has selected PacketType = LoRa. The LR2021 supports multiple modulation types—including LoRa, LR-FHSS, FLRC, FSK/GFSK, OQPSK, and OOK. When using P2P LoRa communication, you must first switch to LoRa mode; all subsequent modulation and packet structure commands will then be interpreted in LoRa mode.
If you select the wrong PacketType (e.g., FLRC or GFSK mode), the chip will not interpret the SF, BW, and CR parameters in LoRa mode, The result is that the transmitted data cannot be decoded by the receiving end at all.
Have you ever wondered what actually happens inside Flutter when you press a button and watch a smooth animation?
Have you ever wondered what actually happens inside Flutter when you press a button and watch a smooth animation?
rexglue (https://habr.com/ru/articles/1002640/
example project with dolphin https://github.com/rexglue/demo-iruka
Replace with our .xex file (project), then start codegen
There’s a dangerous illusion that “vanilla” open source is a silver bullet for enterprise systems. A real-world stress test of recent years has shown the opposite: when familiar giants like Oracle step aside, plain PostgreSQL often turns into a pumpkin under true enterprise workloads.
Mark Rivkin, Head of technical consulting at Postgres Professional, shares his personal perspective on why teams end up reinventing the wheel — adding millions of lines of code to the core — and why the future belongs to converged database systems.
Disclaimer: this article reflects the author’s independent expert opinion.
Continuing the series of CommitFest 19 reviews, today we're covering the January 2026 CommitFest.
The highlights from previous CommitFests are available here: 2025-07, 2025-09, 2025-11.
Partitioning: merging and splitting partitions
pg_dump[all]/pg_restore: dumping and restoring extended statistics
file_fdw: skipping initial rows
Logical replication: enabling and disabling WAL logical decoding without server restart
Monitoring logical replication slot synchronization delays
pg_available_extensions shows extension installation directories
New function pg_get_multixact_stats: multixact usage statistics
Improvements to vacuum and analyze progress monitoring
Vacuum: memory usage information
vacuumdb --dry-run
jsonb_agg optimization
LISTEN/NOTIFY optimization
ICU: character conversion function optimization
The parameter standard_conforming_strings can no longer be disabled
AI video generators in 2026 allow anyone to create high-quality videos from text prompts, images, or scripts in just minutes.
This guide explains how the technology works, compares the leading tools on the market, and highlights their strengths, limitations, and best use cases to help you choose the right solution for your creative or business needs.
After the unexpected divorce between LeCun and Meta, there is a lot of talk that the dead-end in LLM progress will be overcome through the physics of the world. That is, having a neural network work with physical data from the surrounding environment will allow the model to acquire meaning and an understanding of its actions. LeCun has a foundational paper that nobody is going to read. So, I'll summarize it as best I can. Essentially, the idea is that the current trajectory of LLM development is doomed. As long as they are predicting the next token, real understanding — the emergence of real meaning — is impossible. LeCun proposes training neural networks on physical world data, assuming that building a model of it will allow the system to discard details and focus on meaning.
I agree with LeCun that using world data will partially solve the data scarcity problem. But here I see a problem that engineers might not understand. A physical model of the world is actually much poorer than human knowledge. Newton described the entire infinite number of possible falls with a few lines of formulas. I doubt LeCun wants to spend billions of dollars on this wonderful deduction.
This article reviews the November 2025 CommitFest.
For the highlights of the previous two CommitFests, check out our last posts: 2025-07, 2025-09.
Planner: eager aggregation
Converting COUNT(1) and COUNT(not_null_col) to COUNT(*)
Parallel TID Range Scan
COPY … TO with partitioned tables
New function error_on_null
Planner support functions for optimizing set-returning functions (SRF)
SQL-standard style functions with temporary objects
BRIN indexes: using the read stream interface for vacuuming
WAIT FOR: waiting for synchronization between replica and primary
Logical replication of sequences
pg_stat_replication_slots: a counter for memory limit exceeds during logical decoding
pg_buffercache: buffer distribution across OS pages
pg_buffercache: marking buffers as dirty
Statistics reset time for individual relations and functions
Monitoring the volume of full page images written to WAL
New parameter log_autoanalyze_min_duration
psql: search path in the prompt
psql: displaying boolean values
pg_rewind: skip copying WAL segments already present on the target server
pgbench: continue running after SQL command errors
In the IT crowd, it’s common to mock HR for their weird terminology and attempts to assess a “rich inner world” instead of clean code. But when a senior starts tearing juniors apart in code reviews to the point where they end up crying in the bathroom, nobody’s laughing anymore.
We pit two archetypes against each other: the “toxic genius” and the “human glue”. Which one is dead weight, and which one is the load-bearing structure of the project? You might not like the answer.
Hey, Habr! Ivan Glinkin is here again, head of the hardware research group from the Bastion team.
"A flash drive with a combination lock," "a flash drive with hardware encryption," "an encrypted USB drive," and finally, the proper term — "Cryptographic Module". An encrypted USB flash drive goes by many names, but the core concept remains the same.
The purpose of such a device is to protect sensitive information from unauthorized access at both the software and hardware levels through encryption, anti-tampering mechanisms, and various other safeguards. But are these secure USB drives really as reliable as they're made out to be, or is it all just smoke and mirrors?
We decided to look past the marketing claims and conduct our own investigation, attempting to crack several of these devices using hardware reverse engineering. We attempted to extract data, identify the encryption algorithms used, physically open the drives, and read their memory chips.
The results were quite interesting. Read on for the details.
Imagine having a personal AI agent running on your computer. It can read files, run commands, automate tasks, and remember your workflows. In this guide, you will learn how to run OpenClaw with Ollama locally and choose the best local LLM models.
This article is not about Puppeteer being a bad tool. Puppeteer is excellent. And competent TLS fingerprinting will bypass most defenses. But there is a class of tasks where even a perfect network stack won't save you — because detection has long since landed at the level of rendering engine behavior. Let's take a look at how Cloudflare and Akamai expose you through WebGL and Canvas, and why “clean” code no longer works.
There are things in the IT industry whose even existence has become a beautiful myth.
The knowledge described in this article is extremely rare, as it has previously been held by individuals with an academic degree, special training, and, most importantly, access to the necessary equipment.
I recorded a short video showing how to build a complete feature using Claude Code Agent Teams. Even if you're not using Agent Teams, the general approach still applies, making this a useful watch for anyone getting started with Claude Code or curious about this new experimental feature. In the video, I set up an "umbrella" project that brings together two repositories and multiple components: a Java App built with Google ADK (Spring Boot), a Front End App, and a Command Line utility. As part of the demo, the utility is extended with an "api" command that exposes it via the AG-UI protocol, allowing clients to connect to it directly.
Link to the video https://youtu.be/FekjxvNSpyg in the case embed does not work.
Since there are so many great command line AI assistants, including the marvellous Opencode, I thought I was going to stop developing Gaunt Sloth. However, because I control the code and it is so small and LangChain.js-based, I still find myself reaching for it regularly for quick prototypes and other tasks.
https://github.com/Galvanized-Pukeko/gaunt-sloth-assistant/
The last thing I needed was to use it with a very minimal system prompt and without any vendor-provided defaults, so here comes a new config parameter: noDefaultPrompts. This enforces no fallback to default system prompts, which means if your project config does not define a system prompt, it will be completely blank.
Example config
{
"llm": {
"type": "groq",
"model": "openai/gpt-oss-120b"
},
"noDefaultPrompts": true
}
Another new feature, introduced for experiments, is the api command, which currently supports the AG-UI protocol. Calling gsloth api ag-ui will make Gaunt Sloth listen for incoming connections.
{
"llm": {
"type": "openai",
"model": "gpt-4o-mini"
},
"streamOutput": true,
"commands": {
"api": {
"port": 3000,
"cors": {
"allowOrigin": "http://localhost:5555",
"allowMethods": "POST, GET, OPTIONS",
"allowHeaders": "Content-Type, Accept"
}
}
}
}A client implementation and demo setup are available at the link below. https://github.com/Galvanized-Pukeko/galvanized-pukeko-ai-ui
How we transformed stressful manual releases into a dependable, one-click process using GitOps and automation. 50+ modules, auditors and regulators—in a single template that scaled across more than 30 services. No magic, an engineering discipline.
Hello, fellow developers!
Anyone who designs multilayer printed circuit boards in KiCad has faced the need, during routing, to leave large areas free of copper polygons (whether connected to nets or not) for schematic or other reasons. On outer layers, this isn't a big problem. On inner layers, it's a bit different. While it might be fine from a topology perspective, it's not ideal from a manufacturing point of view.
Today, we want to share a plugin we developed to solve this problem. It's a tool for automatically filling free areas on a PCB with non-current-carrying copper elements, either square or round, of a configurable size.
Why is this needed?
At first glance, an empty area on a board is just bare laminate without copper. But for the manufacturer and the end-user of the device, this has two important consequences.