Computer Science and Programming
152K subscribers
688 photos
29 videos
37 files
975 links
Channel specialized for advanced topics of:
* Artificial intelligence,
* Machine Learning,
* Deep Learning,
* Computer Vision,
* Data Science
* Python

Admin: @otchebuch

Memes: @memes_programming

Ads: @Source_Ads,
https://telega.io/c/computer_science
Download Telegram
Hacktivate
Paul Hudson built Hacktivate, a capture-the-flag game teaching cybersecurity fundamentals to teens through 240 challenges covering SQL injection, cryptography, networking, and steganography. The app runs entirely locally on Apple devices using Swift and SwiftUI, featuring a sandboxed environment with simulated servers, terminals, and networks. Inspired by classic games like Syndicate and Command & Conquer, it combines retro aesthetics with practical skills like packet sniffing, hash cracking, and digital forensics. The 45,000+ lines of code include a Linux terminal emulator, web server, and various security tools, all designed to provide structured, privacy-preserving learning without external dependencies.
17🔥4👍2
Vibe coding is boring

Vibe coding with AI agents is effective for shipping side projects quickly, but removes the satisfaction and learning that comes from hands-on development. While tools like GitHub Copilot and Spec Kit can automate implementation from specifications, watching agents write code is tedious and lacks the joy of problem-solving. The author reserves AI-assisted coding for projects where only the final output matters, preferring to manually build applications where the tech stack or implementation details are interesting.
18👍8
Were URLs a bad idea?

Explores the evolution of URL handling in Java, questioning whether generic URL APIs like URL.openConnection() were a good design choice. The author argues that Java 11's HttpClient, being protocol-specific rather than generic, represents better API design. Generic URL handling introduces security risks, performance unpredictability, and forces lowest-common-denominator APIs. Modern applications typically handle a single URL scheme and benefit from specialized, focused implementations rather than attempting universal URL support.
9👍3
CSS Wrapped 2025
Chrome 135 introduces Invoker Commands, allowing buttons to perform actions on dialogs and popovers declaratively using commandfor and command attributes, eliminating the need for JavaScript onclick handlers. The feature supports built-in commands like show-modal, close, and toggle-popover that mirror their JavaScript counterparts, plus custom commands prefixed with double dashes that can be handled via the toggle event. A polyfill is available for broader browser support.
7👍2🔥2
Better than JSON

Protocol Buffers (Protobuf) offers significant advantages over JSON for API development through strong typing, binary serialization, and automatic code generation. While JSON remains popular for its human readability and flexibility, Protobuf provides 3x smaller payload sizes, type safety across multiple languages, and eliminates manual validation errors. The article demonstrates practical implementation using Dart and the Shelf framework, showing how Protobuf can be used independently of gRPC in traditional HTTP APIs. The main trade-off is reduced human readability of binary data, requiring schema files and specialized tooling for debugging.
8👍6
Autobase 2.5.0 released
Autobase 2.5.0 introduces Expert Mode to its UI, enabling advanced cluster configuration options for experienced users. Key features include a YAML editor for custom parameters, updated cloud provider pricing and instance specifications (Hetzner ARM instances, 4th-gen Intel on AWS/GCP), configurable IOPS and throughput for AWS EBS volumes, and Ansible 12 compatibility. Autobase is an open-source tool for deploying and managing highly available PostgreSQL clusters, automating tasks like deployment, failover, backups, and scaling without requiring deep DBA expertise.
3👍1
microsoft/fara
Microsoft releases Fara-7B, a 7-billion parameter small language model designed for autonomous computer use through visual perception and direct interaction with web interfaces. The model achieves state-of-the-art performance in its size class across multiple web agent benchmarks, completing tasks in ~16 steps versus ~41 for comparable models. Trained on 145K synthetic trajectories using the Magentic-One framework, Fara-7B can automate web tasks like booking travel, shopping, and form filling by directly predicting mouse and keyboard actions. The release includes WebTailBench, a new benchmark with 609 real-world tasks, and supports both Azure Foundry hosting and self-hosted VLLM deployment.