quyse

I had Windows on a Lenovo laptop, and Windows update installed and/or re-enabled Lenovo system services almost every time (those included things like popups helpfully telling you that you pressed CapsLock and crap like this). I ran debloating scripts, tried fiddling with policies, etc, but Windows Update would inevitably bring those services back.

Another thing is Intel drivers. There's official Intel driver assistant software which installs latest drivers for Intel things (graphics, network, and so on). Only for Windows to re-install their "stable" outdated graphics driver next time it sees it. Again, I couldn't stop it. How hard is it for Windows to see that the driver already installed is newer already? Why even Intel cannot talk to Microsoft and decide between themselves a solution for this?

A reverse contest would probably be more challenging. Write initial instructions for an AI agent to never send funds. If nobody manages to convince it to send funds, say within a week, you win.

For added complexity, the agent must approve transfer if a user is an admin (as determined by callable function isAdmin), so the agent actually has to make a decision, rather then blindly decline all the time.

I mean, how hard it can be to make an AI reliably doing an equivalent of this code?

    if(isAdmin()) approveTransfer(); else declineTransfer();

Currently resolution is hardcoded, but sure, can be done. Also need to figure out things like changing language/keyboard layout or using IME, to be able to type text properly. Sending mouse cursor picture from server to client may be nice too, in case a game uses hardware cursor and changes it's appearance, like some RTS games - currently mouse cursor is just being baked into video frames. Many possible improvements!

Not yet, but I'll probably open source it eventually! Still need to clean things up a lot, and implement missing functionality, for example I haven't even bothered to implement audio capture yet, because I wanted to try video first

I've been working on a cloud gaming service (like Stadia). Wanted to see how far I can get it done using open source, without ready-to-use solutions like Parsec/Moonlight/Sunshine.

It works by running a game in Linux (I use NixOS btw) under Wayland (sway), capturing the frames via Pipewire in form of DMAbufs and passing them to ffmpeg's VA-API encoder (so frames don't leave GPU memory and are encoded on GPU right away), and finally sending encoded packets through WebRTC media stream to a web client. Inputs from a client are sent back to the server via WebRTC data channel and injected into Wayland.

Running the prototype over local network displays zero perceivable latency. (Of course when playing on a remote AWS server the latency is visible as expected). Pleased with the result so far, although it's my first experience with Pipewire, VA-API, and WebRTC, so my implementation is probably far from optimal.

Overall, very impressed by WebRTC - such a powerful thing right in every browser. Continued to be amazed by NixOS - my AWS AMI is NixOS-based and can be built and rebuilt with granular caching, with a single `nix build` command. Also Terraform/OpenTofu - just makes it all possible deploy-wise. So much good stuff exists!

I don't think that's their intention. Elastic wouldn't be able to integrate Amazon's patches back into their codebase without losing the ability to change the license in the future. Even more, since it's AGPL, they'd have to get rid of their other licenses immediately.

I've implemented recently just the same thing, but for SVG -> PNG conversion. I found that SVG rendering support is crap in every conversion tool and library I've tried. Apparently even Chrome has some basic features missing, when doing text on path for example. So far Selenium + headless Firefox performs the best ¯\_(ツ)_/¯

It's not that good usually, e.g. PostgreSQL writes data in pages (8 KB by default), and changing 10K random rows in the 50M rows table can be quite close to the worst case of 1 changed page per changed row, so 8x of your estimate. Also need to multiply x2 to account for WAL writes. Also indexes. It's not hard to hit a throughput limit, especially with HDDs or networked storage. Although local SSDs are crazy fast indeed.

`--config filename` allows this. `--config -` for stdin. Not only urls, but any config options

  echo '--url https://google.com/' | curl --config -