grrowl

Somehow gemini-cli is so janky and doesn't get the latest models. But Antigravity is the new golden child? Just have a great suite of predictably quality products.

It's the vercel way. There have been plenty of experiments leading up to this (even by vercel employees before they joined) but re-packaging it as "the" solution, rather than just a tool renderer from props (tool schema)

it's back again!

OpenAI seems to limit how "hard" your gpt-5-codex can think depending on your subscription plan; whereas Anthropic/Claude only limits how much use you get. I evaluate Codex every month or so with a problem suited to it, but rarely gets merged over a version produced by Charlie (which yes is $500/mo, but rarely causes problems) or something Claude did in a managed or unmanaged session. ymmv

It's a really good platform for Typescript microservices which scale-to-zero (up to very high theoretical limits), but it wouldn't be a platform you'd migrate a monolith PHP app to (for example).

You're describing why reasoning is such a big deal. It can do this freakout in a safe, internal environment, and once it's recent output is confident enough flip into the "actual output" mode.

This is just an ad for generalanalysis (itself an MCP tool).

Look into zgen, it's a zsh plugin system but much more static (but less ridiculous out of the box)

Great to see Huggingface stick to their guns with CodeEval and python tooling. Agentic turn-by-turn tool calling is fine and all, but we're underutilising their ability to write an execute code in an "agent-like" environment.

Easy to setup and instant forks for devs was it for me. Felt the pain with dev app state in the past and this took <20 minutes to migrate (beta users only) and go live, and get back to feature work.

I remember back in the day you could embed <img src="http://someothersite.com/forum/ucp.php?mode=logout"> in your forum signature and screw with everyone's sessions across the web

Or more recently, NPM's infamous "Progress bar noticeably slows down npm install #11283" issue

[1] https://github.com/npm/npm/issues/11283

My grandparents were terrible at smartphones. To them, it's like a landline phone, but you have to charge it every 1-2 days. Yet my grandpa was decent at PC, and email, and so on, as it was in "a place" and easy to drive.

You can "self-host" as in `next build` and `next serve`, but it runs as a monolothic runtime. Next-on-vercel compiles each api route to a serverless function and supports various additional flavours of server side rendering of pages. Open Next is being able to run like Next-on-vercel but on AWS, your own infra, Cloudflare Workers, and others.

Sounds a lot like existing Claude integration + Caching (as most queries rely on relatively static code context)

It is a native app, and even lives in your menu bar for some reason

Truly, (for those who don't want to log in) it's just a shared playground conversation with GPT-4, in Danish, where the author got a result they're obviously unhappy with.

Author, do some more robust tests, write a blogpost about it, and submit that instead.

it's a "fundamental transformation of the OS itself", you see

It's a nice tip (the existence of URL.parse) but doesn't scream "get shit done" like `import { parseUrl } from 'lib/url'` (or parseJson, etc)

Pumping up the grandiose factor; a meter is tangible and can be imagined 100x, but football fields are really big!

At a certain level you can't guarantee the outcome of calculations will be 1:1. For example python and node both perform floating point calculations differently, so "identical" simulations will drift apart over time.

Mapping and modding Quake 2 and 3 pushed me from Basic and HTML/CSS into much less forgiving C/C++ programming around the same age -- wouldn't be the engineer I am without it. It's invaluable to modify something you know and understand and extremely rewarding, which is the first thing I teach to friends who want to learn.

Can you deploy 700MB Dockerfiles to Cloudflare, as they mention as a minimum requirement in the article?

Worth mentioning Raspberry Pis are designed, and in most cases built and assembled, in the UK.

Article is really just data from their service, and gets it a bit inside-out. Anecdotally, PRs with "!" in the title are often actionable and mergeable, with reviewers pre-involved and ready to merge because of the content, not for the title.

Azure-hosted GPT already lets you "upload your own documents" in their playground; it seems to be similar to how ChatGPT GPT-4 Code Interpreter handles file uploads.

If we switch out LLMs for LZMA algorithms your reasoning still makes sense. Which is a real problem with your reasoning.

Would be worried that the old images will still be distributed, but if an open source maintainer needed to push an update (e.g. for security reasons) they'd need to sign up to the paid account

This might be a bit of an assumption, but it seems likely Meta is willing to lose more on lawyers than you'd be willing to ever spend.

I would hope (but I do not expect) Western Digital to recover the data for the end users. This is a really bad situation for some.