Very obviously vibe coded, but seems quite useful. Only criticism I would have is that the UI, while fancy, is way too busy, and something more simple would serve it much better.

It was kinda long so i just used ai to summarize it _/j

but seriously, wow, it is exactly what I’ve been thinking would be the perfect policy, they really nailed it.

Oh awesome, I had misunderstood that paragraph as it being a VM solution like Winboat, but your clarification makes the project much more exciting

I use a instance running continuwuity and every new/experimental (i.e, matrix 2.0) feature in the client works

From the site:

How does it work?

A virtual machine boots a guest operating system. Compatibility runtimes such as Wine instead implement the Windows interfaces an application needs.

Raven runs the .exe directly on the CPU in a supervised guest process. Some Windows functions execute inside that process, in Raven’s guest components or libraries. Requests for supervisor-managed services cross into the runtime outside it, which manages processes, handles, files and registry state and mediates access to the Linux host.

Raven implements Windows Installer custom actions, side-by-side assemblies through WinSxS and a service manager. These are among the interfaces real installers depend on; application compatibility still needs testing.

The official redistributables can be installed with their classic installers, like you would on Windows (Visual C++ 2005 to 2022, DirectX, .NET Framework 3.5 and 4.8, XNA and OpenAL). Raven protects its own system libraries during installation. Snapshot creation can fail, so check that a checkpoint exists before relying on rollback. Tested applications can use the vendor’s installer; other applications may still fail to install or start.

Graphics go straight to Wayland and Vulkan, with DXVK and vkd3d-proton for Direct3D.

I think the point is to have a system in which the simplicity is enforced, kind of like how some cafés ban electronic devices.

So is bottles next dead then?

Yup, totally different, continuwuity and synapse are the main two, and although there are others, they’re just less good.

The commercial implementation is also open source, but there is a a proprietary version aimed at large enterprise customers that adds better optimization for very large deployments, and better auditing and moderation tools.

See here for all the servers, clients and bridges can also be viewed from there: https://matrix.org/ecosystem/servers/

I’m a little ways into the book, and it’s really good. She might be a horrible person but the memoir is a really good window into a way of thinking.

It’s not amazing, but it’s certainly better than teams.

Unfortunately I have yet to see a platform that works better than discord.

Ah, well idk, I suppose you could try using something like APKmirror

Pretty good icon pack: https://play.google.com/store/apps/details?id=com.nothing.icon.free&hl=en-US

Not really, it’s like building your own furniture, cool, but far from practical and not worth it for most people.

The only way it would help the cause is by decentralising a service that’s mostly on one host, like bluesky, but that’s not a problem on lemmy, and the main advantage can be done by switching to a more niche instance (lemmy.world -> lemdro.id / bsky.app -> eurosky)

Their “AI platform” thing is mostly marketing nonsense, they were around before the AI bubble, it’s the same tech as google translate (technically AI, but not the same kind).

My point was mostly that they should write a text in chinese the way they want then machine translate it to english, so the structure and substance is human-made.

it seems like their first language is chinese, and they can’t write well in english

Just so you know in the future it’d probably be better to write your post in chinese and then use deepl to translate it if you can’t write a good text in english. Might end up a tad stilted but people here would like it a lot more, since the text AI generates in english is both artificial (nobody talks like that) and annoying to read.

I may be misinformed but the article says

We’ve verified that every AST produced by the Rust parser is identical to the C++ one, and all bytecode generated by the Rust compiler is identical to the C++ compiler’s output.

which I took to mean that the new code behaves 100% identically to the old code, like a byte-perfect decompilation. If that’s not the case and it’s just the same in their suite of tests then I’d agree with you.

Yeah, using AI to help you rewrite code (which is tedious) in a scenario where there are clear test cases (it either works or it doesn’t) is basically the ideal usecase.

Actually not a terrible way to use AI. Hopefully the port goes well