Taking a look at compression algorithms

Taking a look at compression algorithms
35 minutes by Moncef Abboud

Moncef built his own Kafka implementation called MonKafka and dove deep into compression algorithms. He explored how compression works by reducing data redundancy to save storage and transmission costs. The post covers four main algorithms: GZIP uses DEFLATE combining LZ77 backreferences with Huffman encoding, Snappy focuses on speed over compression ratio, LZ4 is even faster with similar ratios to Snappy, and ZSTD achieves the best balance with high compression ratios at fast speeds.

Introducing world's first Generative UI API – Don’t Ship AI apps with legacy UI
sponsored by Thesys

C1 is a Generative UI API that augments LLMs to respond with interactive UI like charts, forms, cards, and more in real-time, instead of plain text. It integrates seamlessly with any LLM, framework, or MCP in just 2 lines of code. C1 renders on-brand UI with customizable components and consistent behavior via system prompts. Ship rich, interactive AI apps 10× faster with 80% less UI overhead.

How AWS S3 serves over petabyte per second with slow HDDs
11 minutes by Stanislav Kozlovski

S3 stores over 400 trillion objects and handles 150 million requests per second using cheap hard drives instead of expensive SSDs. The system overcomes the slow random access speeds of hard drives through massive parallelism. It splits files into pieces using erasure coding and spreads them across thousands of drives, allowing parallel reads that achieve much higher throughput than any single drive could provide.

Address matching using a fault tolerant trie
2 minutes by Robin Linacre

A trie data structure can match messy addresses to clean ones by using the natural hierarchy of address parts. It stores addresses as a tree where each node represents a word and tracks useful information like whether it ends an address. The system becomes powerful when fault tolerance rules are added. These rules allow skipping extra words at the start, middle, or end of messy addresses to find matches in the clean address database.

Programming deflation
4 minutes by Kent Beck

AI tools are making coding cheaper and faster, creating a paradox. Economic theory suggests this could either reduce the need for programmers or increase demand for programming. Unlike traditional deflation, this change comes from genuine productivity gains rather than economic weakness. As basic coding becomes commoditized, value shifts to understanding what to build and integrating complex systems. The key is developing skills like judgment and system thinking that remain valuable regardless of whether we get fewer or more programmers.

How Obsidian reduces the risk of supply chain attacks
2 minutes by Lishid

Obsidian protects against supply chain attacks by keeping dependencies to a minimum and building features from scratch when possible. The team strictly version-pins all dependencies, avoids automatic scripts, and takes time to thoroughly review any updates before releasing them. This slow, careful approach creates a safety buffer that helps catch malicious code before it reaches users.

And the most popular article from the last issue was:

• How software engineers make productive decisions