ObjC/Swift have ARC, etc. But, it's fairly easy (my opinion) to accidentally hold onto references. This isn't a garbage collection issue. It's an object lifetime issue. Not being pedantic. Commiserating.

dear algo 👋 connect me with people who love: -💻 retro sysadmin to modern day ai dev -🖥️ mac osx linux ios droid windows atari -⌨️ bash c c# objc java go rust py php next -🛠️ infrastructure -🤗 experience -🏗️ content 👀 -🤖 the ai ml maths physics rabbit hole ty >> b0b 🙌

real native objc

Do you prefer swift or objc? I tried to look at swift but it look much more complicated and turned me away. I kind of like the simplicity of ObjC. From this wwdc I kind of got the impression the ObjC is probably going to be nuked in some years.

Your AI coding agent has a dirty secret: It doesn't understand your codebase. It greps it. Over and over. Burning your tokens on every search. Watch Claude Code on a big repo — it spawns Explore agents that grep, glob, and Read files in loops just to figure out what calls what. You're paying for an LLM to do what a database does for free. CodeGraph fixes this at the root. It builds a local knowledge graph of your entire repo — every symbol, call graph, route, and dependency, across 20 languages — and plugs it into your agent as an MCP server. The agent stops searching and starts querying. One tool call. Instant answer. Even across language boundaries grep can't follow — Swift ↔ ObjC bridges, React Native modules, dynamic dispatch. Benchmarked on real codebases — VS Code, Django, Tokio, OkHttp, Excalidraw: → Up to 81% fewer tool calls (VS Code) → Up to 64% fewer tokens → Up to 40% cheaper (Alamofire) → Faster on every single repo tested Average across 7 languages: 47% fewer tokens, 58% fewer tool calls, 22% faster. Works with Claude Code, Cursor, Codex, OpenCode, Gemini CLI, Antigravity, Kiro, Hermes. Setup: npx @colbymchenry/codegraph → 100% local. SQLite only. → No API keys. No cloud. Nothing leaves your machine. → Auto-syncs as you code. Zero config — literally no config file exists. 47k stars and climbing. MIT licensed. The next leap in AI coding isn't a smarter model. It's stopping the model from wasting its intelligence on file search. Link in comments 👇

Rustで遊んでないでもっと早くObjCで作っとけよ。

Literally I lost interview coding round writing ✍️ objc code. I changed since then till swift arrived to learn python just to make through coding rounds

听起来 macOS 27 和 iOS 27 本质上就是：苹果把积累了几十年的 ObjC Swift 屎山代码库扔给 Claude 扫雷，修出一堆从来没人发现的 Bug，然后把修复记录包装成 WWDC 发布会内容。

TIL the objc selector base in the dyld shared cache points at 🤯 a string that can never be a real selector, so that reading it back confirms your offset math, cool

Yeah, but we can just make some header or modulemap to use it since most of the API is ObjC.

Ще трохи бріджу swift-objc-c

Dh on objc xzcohxic9jh0k.g mo. Hzch hh j .n o e .

I loved objc. It had its own purposes, and for them, it was great!

The C/C /ObjC separate compilation architecture is one of the most underappreciated PL innovations of all time

objc uikit forever