Hacker Newsroom for 04 May: Mercedes Physical Buttons, Ladybird April Update, watchOS Maps Journey, Mercury Haskell Scale

Hacker Newsroom for 04 May recaps major Hacker News stories, moving through mercedes physical buttons, ladybird april update, watchos maps journey, mercury haskell scale. 1. Mercedes Physical Buttons The next story is about Mercedes-Benz bringing back physical buttons in its upcoming interiors after customers told the company that touch-sensitive controls and buried menus were frustrating. The article says Mercedes will keep large screens, including the big MBUX Hyperscreen, but add hard keys for key functions on the new GLC and C-Class, along with buttons and switches on the steering wheel. Story link Hacker News discussion 2. Ladybird April Update The next story is Ladybird's April 2026 project update, and the post says the browser merged 333 pull requests from 35 contributors while moving a lot closer to everyday use. The big changes include inline PDF viewing, richer history-aware address bar suggestions, incremental and speculative HTML parsing, off-thread JavaScript compilation, independent rasterization for each navigable, a new GTK4 frontend, and a long list of CSS, networking, and performance fixes. Story link Hacker News discussion 3. watchOS Maps Journey The next story is Six years perfecting maps on watchOS, a post about how Pedometer++ evolved from early server-rendered maps to a SwiftUI-native map engine, a custom basemap, and a watch layout that finally makes wrist navigation feel polished and practical. The article is really about steady product refinement: shipping better mapping, tighter design constraints, and a more useful outdoor experience on a tiny screen. Story link Hacker News discussion 4. Mercury Haskell Scale The next story is a Haskell blog post about how Mercury runs roughly 2 million lines of Haskell in production while handling serious banking workloads. The article argues that Haskell works there because it turns operational knowledge into types, keeps dangerous behavior behind tight boundaries, and makes the safe path the easy path for a fast-changing team. Story link Hacker News discussion 5. Kimi Coding Benchmark The next story is a ThinkPol news story about an AI coding contest where Kimi K2. 6, an open-weights model from Moonshot AI, beat Claude, GPT-5. Story link Hacker News discussion 6. OpenAI ER Triage The next story is a Guardian report on a Harvard trial that found OpenAI's o1 diagnosed emergency-room cases more accurately than triage doctors when both were given the same text-based records. The article says the model got the exact or close diagnosis in 67 percent of cases, improved further with more detail, and even outperformed doctors on some treatment-planning tasks, but the researchers stressed it still looks more like a second-opinion tool than a replacement. Story link Hacker News discussion That's it for today, I hope this is going to help you build some cool things.