Keeping secrets out of logs

Keeping secrets out of logs
34 minutes by Allan Reyes

Keeping secrets out of logs is both annoying and difficult because there's no single solution to fix the problem. Allan presents 10 "lead bullets" including domain primitives, read-once objects, log formatters, and taint checking. These techniques work together as defense-in-depth rather than standalone fixes.

Building Low Latency Apps Masterclass
sponsored by ScyllaDB

Discover practical techniques for high-performance applications in this free, virtual masterclass led by Pekka Enberg – author of the new book, Latency. Attendees will learn a holistic approach, including techniques that intersect many areas of software engineering, distributed systems, databases, and OSes.

How I solved a distributed queue problem after 15 years
4 minutes by Jeremy Edberg

Reddit used RabbitMQ message queues to handle user actions like upvotes. When users voted, the action went to a queue first, then got processed and written to the database later. This system was scalable and allowed parallel processing, but it often lost data when components failed or crashed. Durable queues solve this problem by saving task progress to a database like Postgres. If something breaks, the system can restart from where it left off instead of losing work. They work better for important tasks but are slower than traditional queues.

Will amazon S3 vectors kill vector databases—or save them?
15 minutes by James Luan

AWS introduced S3 Vectors, a new vector storage solution that stores and queries vector embeddings directly in Amazon S3. The service offers extremely low storage costs at just $0.06 per GB, making it over 10 times cheaper than traditional vector databases. However, it comes with significant limitations including 500-700ms query latency, reduced precision around 85-90%, and caps on collection size and query throughput. James argues this validates the shift toward tiered vector storage rather than replacing specialized vector databases entirely.

Fenwick layout for interval trees
12 minutes by Alisa Sireneva

Researchers combined Fenwick trees and interval trees to create a faster data structure. The new approach uses the unique center points of interval tree nodes as indices, similar to how Fenwick trees work. This eliminates the need for extra pointers and simplifies navigation through the tree. The implementation is more cache-friendly and reduces space usage compared to traditional interval trees.

The story of how RSS beat Microsoft
7 minutes by Ryan Farley

RSS beat the more advanced ICE syndication standard using the same strategy that helped VHS defeat Betamax. While ICE offered sophisticated features for corporate publishers and cost $50,000 per server, RSS remained free, simple, and open to anyone. RSS required only three basic elements to work, compared to ICE's complex 58,000-word setup guide. When even major publishers like The New York Times chose RSS over ICE in 2002, the format war was essentially over.

And the most popular article from the last issue was:

• A history of Chrome