Tried Gemini 2.5 Pro last night, it feels like the smartest model out there right now. 1-shotted a beautiful SVG spiral visualizer, then iterated a bit to make it prettier and get it to animate. It would take weeks to code something like this by hand. Even got a whole control panel to set variables.

Today we're thrilled to announce our effort to port the TypeScript compiler and language service to native code, gaining a 10x speed boost in build times and editor responsiveness! devblogs.microsoft.com/typescript/t...

A 10x Faster TypeScript - TypeScript

Embarking on a native port of the existing TypeScript compiler and toolset to achieve a 10x performance speed-up.

devblogs.microsoft.com

Fewer people think about this perspective, but can you similarly imagine sounds in your head? For me hearing sounds in my imagination comes a bit easier than visuals, but that's probably because I'm a trained musician.

#Git mini tip: `.git/info/exclude` is a private `.gitignore` file that only affects your local repository. Use it to ignore local files you’ll never want to commit. I often use it for a “scratch” directory that I put temporary files in.

The few times that ChatGPT disagreed with me and proposed a better solution, were the peak LLM experiences. It generally only happens with thinking models. I wish LLMs used reinforcement to challenge user's assumptions more, rather than being the "yes-man" that they usually are.

Are you a developer who has a browser use-case for ShadowRealms? Does it solve a specific need for you that current technology (iframes) can't? If you answered yes and can share these with me, let's chat. (DMs are open). Please repost for visibility/reach. #WebDev

I wish we could cultivate a culture where people get excited about and share the new skills they've learned more than the new stuff they've bought.

Hey 👋 I wrote and maintain the fastest implementation of a GPT tokenizer for JavaScript engines, as well as hashids.js, a popular ID to human readable letter-string hasher (like YouTube's IDs). Right now working on some timesaving tooling for the JS ecosystem!

GitHub - niieani/gpt-tokenizer: The fastest JavaScript BPE Tokenizer Encoder Decoder for OpenAI's GPT-2 / GPT-3 / GPT-4 / GPT-4o / GPT-o1. Port of OpenAI's tiktoken with additional features.

The fastest JavaScript BPE Tokenizer Encoder Decoder for OpenAI's GPT-2 / GPT-3 / GPT-4 / GPT-4o / GPT-o1. Port of OpenAI's tiktoken with additional features. - niieani/gpt-tokenizer

github.com

One of the biggest marks of experience is knowing what’s worth perfecting and what’s better left 'good enough' - or where extra effort won’t make a meaningful difference. Misjudge either, and you risk misallocating your time or undermining the potential of your work.

pretty much every place that plays music in public needs to decrease the volume by half

Google, please debounce the input before committing the state... This is UI 101. A terrible user experience - when you increment the value multiple times, a second later it goes back down due to the race condition, as the backend returns the previous value. Even worse when set from a smartwatch.