CIRCLE WITH A DOT

@mcc GrapheneOS will keep all of the junk out for as long as it is able to keep existing.

@ska @dalias @ariadne The real waste is a ton of money being wasted that could have been spent on more useful things. Like solving world hunger.

@ariadne Can it be done? Yes. Is it a good idea? Absolutely not.

The cooling and power problems are absolutely solveable with enough work. The financial problems are not solveable in the forseeable future.

@chris It’s a natural response, but it isn’t always the best one. For instance, allegations of abuse or harrassment need to be taken seriously.

RE: https://eldritch.cafe/@EllisArcwolf/116131859022211863

Seriously, read this post. Then read the blog post it links to.

Ellis has done their research on this, and the reality is vastly more nuanced than most people realize.

@whitequark @navi @SRAZKVT Main problems with Flatpak are:

1. Some upstreams (you almost certainly not included) don’t update dependencies when there are major security vulnerabilities. For instance, OBS Studio shipped an old CEF that had a Chromium version riddled with exploitable holes.
2. It only works (well) for graphical applications. CLI tools need hand-written wrappers, and it doesn’t work for daemons, libraries, or embedded devices.
3. It blocks user namespaces, breaking browser sandboxes. I believe WebKit and Gecko (Firefox) have alternative sandboxing options, but they have more overhead. Chromium doesn’t have an upstream alternative at all, which is unfortunate because it is the most secure browser engine.

@whitequark I think it is also because big corporations are the ones who see the impact of memory unsafety at large scale. Individuals may be aware that the problem exists, but I suspect most aren’t aware of its scale.

@navi @whitequark @SRAZKVT Android dynamically links its Rust code. This does require rebuilding programs when their dependencies change, but for a closed system like Android that isn’t a problem.

@ireneista Static or dynamic?

@dalias Is splice even useful nowadays?

@ireneista Isn’t the complete lack of type checking in Forth already cruel?

@ireneista preadv2, pwritev2, send, recv, sendmsg, recvmsg, and the like all have ways to avoid blocking.

@lina US and EU law seems to be pointing in the direction of the model providers being potentially liable, but model users not being unless they do something stupid (like prompting the model to get those violations out).

@ariadne is this something that could be fixed or is it too fundamental to how Matrix works?