[Dev Catch Up #52] - Microsoft's Majorana-1 QPU, Claude Code and 3.7 Sonnet, Grok-3, debugging Linux hung task, and much more.
Bringing devs up to speed on the latest dev news from the trends including, a bunch of exciting developments and articles.
Welcome to the 52nd edition of DevShorts, Dev Catch Up!
For those who joined recently or are reading Dev Catch Up for the first time, I write about developer stories and open source, partly based on my work and experience interacting with people all over the globe.
Thanks for reading Dev Shorts! Subscribe for free to receive new posts and support my work.
Some recent issues from Dev Catch up:
How to use Postgres as a Vector Database with BGE Embedding model
How to do Local AI Code Generation and Completion using Open-source Llama.cpp?
Join 6500+ developers to hear stories from Open source and technology.
Must Read
Microsoft stunned the tech world with the unveiling of Majorana 1, the world’s first quantum processing unit. I was researching more about it and found out that the topological core that is powering it is designed to scale to a million qubits on a single chip. Learn more about this giant leap in technology from Microsoft’s product announcement article.
Recently, Anthropic introduced Claude 3.7 Sonnet, the most intelligent model of the Claude family similar to o3 but better because it looks and feels like an integrated model. Apart from this, I was more excited about the launch of Claude Code, which is a command line tool for agentic coding. The official Anthropic announcement blog sheds more light on both of them.
Previously in some of our editions, I have covered chain-of-thought reasoning in LLMs. While digging deep into the subject, I found multiple papers covering it but this paper caught my attention. It proposes Hijacking Chain-of-Thought (H-CoT), a universal and transferable attack method that leverages the model's own displayed intermediate reasoning to jailbreak its safety reasoning mechanism.
React has been around for a longtime as a popular frontend-tech choice for developers. To develop large scale application with it, you need to have a knowledge of the many useful libraries. This article covers the essential React libraries for 2025 and helps you navigate through the ecosystem effortlessly.
OSS Highlight of the Week
Ripgrep is a line-oriented search tool that recursively searches the current directory for a regex pattern while respecting your gitignore. It will respect the rules and automatically skip hidden files/directories and binary files. Check it out from its official GitHub page.
Good to know
While LLMs improved significantly with completing complex reasoning tasks, their efficiency is not up to the mark due to substantial memory and computational costs associated with generating lengthy tokens. I discovered a paper that counters this issue by introducing LightThinker, a novel method that enables LLMs to dynamically compress intermediate thoughts during reasoning.
UV, is the missing performant package manager for Python, made by Astral. Here is an article that shares more information on the tool, advises devs on the pros and cons, and gives an opinion on the migration.
Recently, XAI released their most powerful Grok-3 model trained on Colossus supercluster with 10x computational power than the previous one. The XAI official article covers all about the model and claims that the model is performing better than other major models of the same category in different benchmarks. Apart from this researchers like Andrej Karpathy detailed the pros and cons of the model in his explanatory x.com tweet.
If you are a developer, you might be familiar with writing unit tests. Mocking is an essential part of testing and it makes tests more reliable and faster. This article covers Nock, an API mocking library and discusses how it revolutionizes mocking with Node.js.
LLMs are beyond mere chatbots and their improvement have been exponential. But programs submitted to LLM serving engines experience long cumulative wait times. I came across this research paper that aims to mitigate this issue with Autellix, an LLM serving system that treats programs as first-class citizens to minimize their end-to-end latencies.
Network safety is one of the most important security component that an organization can focus on. DNS vulnerability could allow an attacker to redirect network clients to alternate servers of his own choosing. Steve Friedl summed up everything about it with this illustrative guide.
Notable FYIs
The Linux kernel can produce a hung task warning message in its log. Here is an article that covers how hung task warning work, the reason behind their happenings, and much more.
Developing a software is something we all know but there is not much information available on the long-term maintenance of the same. I found this article that covers the perspective of long-term software maintenance from a maintainer’s perspective.
Next.js is a framework that has been loved by all of us. While it has offered a lot of amazing things related to development, there is much more to offer in the future. This article covers everything on the future of the framework.
Another fruitful development happened in the world of AI coding assistants, with Google announcing that Gemini Code Assist is completely free. Learn everything about it from the official Google article.
If you are a C++ dev, this one interests you.
std::generatoris a C++ feature that enables you to write concise, straightforward functions that generate sequences of values on-demand without manually managing state. Here is a guided walkthrough on the usage of the feature with an example.
That’s it from us with this edition. We hope you are going away with a ton of new information. Lastly, share this newsletter with your colleagues and pals if you find it valuable. A subscription to the newsletter will be awesome if you are reading it for the first time.