I don't care if you have 5 followers or 5,000 followers.
I don't care if you get 5 likes or 5,000 likes.
If you are my friend, I'm supporting your tweet.
Given the total lack of respect for open development and integration, we will be limiting our Twitter presence to just release announcements and likes. See you over on 🦣! Can't link to it either because of a ban on cross site promotion!!
arstechnica.com/?p=1911442
I feel that many developers in North America don't appreciate how huge it is when a developer in India saves up for their own MacBook.
The average salary in India is ₹32,840 per month.
The cost of a MacBook Pro M2?
₹199,900
Qwen3 Coder has a 5.32% diff edit failure rate in Cline, based on real-world data.
It's right alongside Claude Sonnet 4 and Kimi K2 as an excellent model in performing diff edits.
Not bad for an open-source model at $0.30/$1.20 per million input/output tokens 👀
China dropped these open-source models in July:
- GLM-4.5
- GLM-4.5-Air
- Wan-2.2
- Qwen3 Coder
- Qwen3-235B-A22B-Thinking-2507
- Qwen3-235B-A22B-2507
- Kimi K2
Meanwhile:
- OpenAI still hasn’t released the open-source model
- Anthropic is doing stricter rate limits
- Meta might
Noticed something cool with Firebase today — it updates one file at a time, waits for your review, then moves on. Feels slower, but way calmer. You actually see what’s changing. Devs, you’ll like it.
ClineRoo CodeKilo — should totally have this. What say?
Started a small Angular project. Coming from React , it’s nice not having to choose routers, state libs, or worry about RSC. Feels good to just build—framework life ain’t bad.
Lesson learned: never let agentic tools modify non-versioned files. Asked an AI agent to add a command-line alias—now my entire .bashrc is messed up! 😩
2 requests for Claude Code's 5-hour limit:
1. At least finish the task or clean up half-written code before cutting off!
2. Let me queue or schedule tasks for it to pick up when the next 5-hour window starts.