Very nice. Macs have a new containers program. It’s open source on github but not ready for the current MacOS version. Might be an even better approach as it’s made by Apple.
I followed his posts internally before he left. He was strict about resource waste. Hand tracking would break constantly and he brought metrics to his posts. His whole point was that Apple has hardware nailed down and it’ll be efficient software that will be the differentiator. The bloat at Meta was the result of empire building.
Opening an animal crossing next to a school in an area where there are many coyotes is a lot different than fingers pinched by doors. But if you’re making the argument that this animal crossing is so important than the occasional child being killed by animals is as trivial as fingers being pinched by doors, I know all I need to know about you.
The real fear I have is in plausible deniability for filtering and muting discussion, which they do now. Moving it to a NN can always hand wave away the responsibility.
You’d be calling libraries that are vetted by the security team. Engineering at Meta is a lot like coloring with crayons. It’s very limiting but the infra does a lot for you.
Does anyone have a good EBNF notation for Sqlite? I tried to make a tree-sitter grammar, which produces C code and great Rust bindings for it. But they use some lemon parser. Not sure how to read the grammar from that.
If you use jupyterlite, you're using the same thing. Bento is just the internal Meta version and the only potential benefits is the internal integration.
Meta doesn't use git. It uses mercurial. It does fork it because they have a huge monorepo. They created a concept of stacked commits which is a way of not having branches. Each commit is in a stack and then merged into master. Lots of things built for scaling.
The author took a tangent so fast and so long into some theater that they didn't ever explain what his problem with AI is. I would love to hear more about it and specifically how Google changed in the last 3 years. I'd like to know what the outlook was 3 years ago and how it drifted.
Code generation is very common at Meta. One just needs to create a query or fragment. A fragment can be spread at the root query level using relay so it will do the fetch once. It gets more complex because you can lazily query more data based on parameters you pass into GQL. It feels like magic and is really annoying imo.
The tweet that is linked in the web page is from the former managing editor of Al Jazeera. You can just click on the tweet directly and use Google translate.
I used Pyodide (https://pyodide.org/en/stable/) as a python execution environment in browser. That was pretty successful. There are also python libraries that let you generate a config which is later made responsive via a JS library. Pyodide runs in wasm.