A small bouncy bald Brit.
Passive Londoner, active Oaklander.
Places I've worked (in reverse order): Honeycomb.io, LaunchDarkly, Compaas, 18F (GSA, US Gov), Neo Innovation, Linden Lab, Ning, BDQ, Sparza, h2g2/The Digital Village, Delphi.
This is probably naive of me, but avoidance of a real sandbox sounds like a great idea so far:
1: Apply the Principle of Least Power/Privilege, especially where agents are involved. "With great power comes astonishing fuckupitude."
2: More freedom to diverge from the existing POSIX and bash primitives to something that better suits the data involved
Having logical control of the filesystem lets you do fun tricks like using file paths as a simplistic search API (e.g. /docs/tag1/tag2/ lists all the docs tagged with both tag1 and tag2).
But it might also be worth looking at shells which give you more power over structured data, like nushell. It's written in Rust, but it might be relatively simple to port just the bits you need to Elixir (which I would never have said a couple of years ago). Keeping everything in process sounds like a major win to me, because of both performance and control.
Yeah, it's a shame. I usually enjoy Raymond Chen's posts, but this doesn't tell me anything particularly interesting other than that the band didn't know. The main question I have is: why this video? What's the story behind the choice?
Dan Davies wrote a whole book on this topic, The Unaccountability Machine [1]. In it, he creates the concept of "accountability sinks": organisational structures or systems which abstract the source of a decision away from individuals so that no specific person can be held accountable.
Davies writes: "For an accountability sink to function, it has to break a link; it has to prevent the feedback of the person affected by the decision from affecting the operation of the system."
"The StandaloneWindow module provides a way to create a separate window to display custom content. You are able to control a full-sized webview inside the window, therefore you can use HTML, CSS, and advanced techniques like WebGL to render the contents." [1]
So... someone could make a Webamp[2] plugin?
And Butterchurn[3] for viz? (Assuming one can plumb in a compatible audio node)
This is amazing, but there are even wilder ways to run arbitrary code inside a PDF. How about stringing together several thousand segment commands in JBIG2 (one of the image codecs supported in PDF) to create a programmable virtual machine? https://googleprojectzero.blogspot.com/2021/12/a-deep-dive-i...
As everyone else has been saying, this is amazing work. It sounds like the biggest issue with loading the page is the initial sql.js download - it's about 1.2MB, is that right?
Might it be feasible to easily strip down SQLite so that it only compiles the parts for read-only use? The browser version is obviously somewhat read-only but that's because of the sandbox. I'm talking about excluding the code for CREATE, UPDATE, INSERT and everything else which is just for writing. The aim here would be to produce a significantly smaller WASM binary.
I'm guessing that the answer is no, there's no easy way of doing this without significant rewrites of SQLite's core, but... I can't be the only one to think of this, surely?
Have you looked at the iOS App Store lately? There are tens of thousands of games there. Apple Arcade has a small (100+) but well-curated selection of good games that run on both iOS and MacOS.
Oh, you meant PC games? Sure, the Mac only has a small percentage of (for example) Steam games, but that percentage is steadily rising - it's now over 25%. Switching architecture is unlikely to present a major problem for most developers, especially given that they're probably using Unity or Unreal Engine.
Places I've worked (in reverse order): Honeycomb.io, LaunchDarkly, Compaas, 18F (GSA, US Gov), Neo Innovation, Linden Lab, Ning, BDQ, Sparza, h2g2/The Digital Village, Delphi.