I have vibe coded a remote X/Wayland solution that actually works really well for this, at least as long as you are mainly interested in one machine (so good for people with remote work machines, not so good for network/infra engineers).
You just run `mycommand connect host` and it will connect via SSH, start a Wayland server with X support, set up the appropriate environment variables, and also automatically reconnect, so you only have to do this once.
After that you can just run GUI commands via SSH and appear locally like you always wished they would. It uses a VNC-like image data protocol so the client works on any OS. And windows are seamless - no windows within windows nonsense. I haven't put much effort into the video protocol tbh but it's already better than Xpra and very usable with 100ms latency (London to NY) so that's good enough for me.
I'll try to open source it.
I think a better solution would be if SSH itself supported some GUI channel, but the only thing is the Kitty graphics protocol which kind of sucks tbh. I did try writing some tools using that for the GUI a while back but it's awkward because you have to use PNG, you're still stuck with the character grid for text, etc.
100ms sounds good, but how about the consistency of the GUI (remote vs local) and HiDPI issues when it comes to this approach? (I assume latency would be higher when there are many applications running in many windows?)
To be clear I meant the experience is fine when running on a 100ms link. I haven't measured the actual latency. That said it uses QUIC and sends the data as tiles, which are displayed immediately, so I suspect it is basically as low latency as possible. It does lead to fun mosaic style tearing which half of me finds cool and half wishes it didn't do that, but side-by-side with Xpra it's much faster and less annoying than waiting for a whole frame which is what Xpra does.
I'm running an old Qt X11 app with a modern retina MacBook with multiple different density screens and it works fine. There were some minor issues with that to start with but I pretty much just said "this happens, please fix it" and Claude figured it out without even having access to a Mac, which is impressive.
Honestly the thing that was hardest was keyboard layout. That kind of stuff never worked well for me in any remote desktop app, and now I can see why! You might never notice if you only use the US layout tbf. It works pretty well now though - I got it to query your local layout and the dynamically build an X keyboard layout and send it to the server.
I've run it with up to about 4 large windows and it seems to work fine. It doesn't send anything when nothing is changing so lots of windows won't matter unless they're playing videos or whatever - it's more designed for old Windows/Qt/Gtk style apps. I suspect you could do better with a traditional video codec but unfortunately the server I use has no GPU so that wasn't really an option.
ssh -X (or -Y) forwards your X11 connection automagically. It doesn't work for Wayland because Wayland was designed to be impossible to forward.
However, most X11 apps will have utterly abysmal performance on WAN SSH connections. Ancient apps will be slow because they block waiting for every response (this is unavoidable in Xlib) and less than ancient apps will be slow because they send full pixel data. It is possible to write a reasonably performant remote X11 app (using XCB instead of Xlib and always being aware of latency), but there are very few such apps. This is why Wayland dropped the requirement.
> With 100% agentic vibecoding, everything should be amazing now.
At least in my company, it is. I have vibe coded three apps that are far better than the competition: an Xpra/Wprs alternative that's faster and much more reliable, an alternative to the VSCode Git Graph extension that is much faster and has tons more useful features (e.g. drag and drop to move/copy/rebase commits which is a feature I always wanted), and a Mac app that lets you use cheap MIDI controllers to control your screen brightness, volume, media playback etc.
I use them all every day and they're all extremely reliable, and much better than the existing tools. One was entirely vibe coded and the other two were unfinished projects that I vibe completed (probably 70% vibe code).
I suspect a mountain of people and companies have tools like these but they haven't open sourced them because a) I used company tokens so there's admin to get things open sourced, and b) it almost feels like there's no point... you can just vibe code your own version to your exact liking. Do you really want to use mine?
One major issue is also, the negative feedback loop on anything coded with a LLM.
The amount of posts when somebody introduces their program here on HN, that may have been LLM coded but that does not neglect the often massive amount of time the "dev" put into it. And yet, its all dismissed under this single negative "vibecoded" label.
I noticed that we are seeing more and more closed source programs, because people do not want to open source LLM generated code, out of fear that people start a negative PR loop against their program. And this reflects back to people not even releasing their programs into the wild.
There seem to be a inability of people to tell the difference between somebody spending a few prompts on making something, vs somebody spending months on 10.000's of prompts to make a great working program. It all gets thrown under that "vibe coded" bus. When its ironic because most code by now is LLM coded anyway. The amount of people hand coding is now probably a absolute minority.
I don't want to open source my code because thievery and entitlement and generally unfriendly culture towards one another, anti to the original spirit of open source, has become more prevalent.
how/why that has happened is another story. I don't see the value in giving away my valuable thinking and ideas when it isn't appreciated and this massive-theft-founded AI company culture of our time thinks it is entitled to everything we create.
[Controversial edit: I don't think of myself as a slave, no matter what others may try to convince me to believe.]
Well, I think that's a bit of a shame but I can also come up with reasonable explanations for it...
Mostly, I think if you are promoting outside of your main area of expertise and couldn't have written it on your own, people don't want to put stuff out there and portray themselves as an expert in a subject matter that they relied heavily on AI assistance for, because it wouldn't be honest to other contributors to said project or other similar projects that were done without AI assistance. It would be a bit of a false representation of your own skills. That's why I am all for publishing, but also all for disclosures of AI assistance.
So, I think I'd be fine with taking a look at and contributing to "vibe coded" projects so long as the project leader disclosed their level of expertise on the underlying subject matter that the repo maintainer has and the extent of the AI assistance.
Then it may be a genuine learning opportunity for them (and for me) and more power to them and it's a shame if it doesn't get published, but I can see why socially there is a pressure not to do it, and get ragged on by the actual SMEs in the area that no doubt will be able to poke holes in what the AI came up with. I think we need those types of collabs though to upskill these new AI assisted coders into difficult types of programming though, so like I said, a bit of a shame if it never gets published and nobody that is an actual SME gives it the time of day.
I think that's actually a reason behind why the Linux Kernel has made decisions to allow AI assisted contributions by the way. I think they are trying to turn some of these people that want to be systems programmers (but at the moment can only do so through AI assistance) into real systems programmers by getting them involved and getting them mentorship. Which is I think a good thing. As long as Linus and other SMEs are there to keep a handle on potential actual AI slop, then it represents an opportunity to upskill interested people into a very not easy kind of programming. They just need to be treated fairly and disclose and get the mentorship they need to develop what I can only call "taste".
I agree with him. I only managed to get asterisks enabled by default in sudo because sudo-rs isn't owned by people who have decided that the 70s was the perfect decade and nothing should ever change. And look at the insane pushback when it did!
I can't even get an obvious UX improvement through PAM because they like it broken.
I wonder when we will move onto harder tasks -- I remember around 2007 people were using Crysis (also a first-person shooter) as a test of new computers as when it was released hardly any computer could run it without turning down all the settings. These days I'd expect most (although not all) low end hardware could run that well.
Must have been on the first attempts, right? As the name suggests, a meme (mimesis) is about imitation but also propagation at scale, otherwise everything would be a meme. The first attempts count as original. Once it gets traction others could very well be genuine technical achievements but in an "also ran" format.
It's like a series where every episode follows the exact same blueprint with very minor changes to the details. It doesn't get old from the first couple of episodes, but it sure does in a few more.
And I was thinking exactly of the "Doom on pregnancy test" example myself but that's worse than average. If my memory serves me, that was a case of using just the plastic shell of a pregnancy test and replacing all the hardware.
> > and agents don't report bugs, they just work it around somehow.
Yeah I've definitely noticed this. On a project I'm working on Opus 5.5 found a bug in Smithay and another in Winit. It did tell me about them but its preferred solution was just to work around them locally. To be fair we don't really want agents going off and reporting bugs automatically...
I don't run any big open source projects but I think my preferred solution to AI PRs would be to get my AI to implement the same thing, and give then afterwards tell it to look at the PR for anything it missed. Then I at least know that I'm using a competent model and it reduces the danger of just blindly accepting someone else's code. (And obviously review the code if you have time and care, but that isn't always the case.)
Yes the default behavior of working around the bug is good, that’s what everyone was doing anyway pre-LLMs. The good practice is to report the bug. Simply add instructions to your agents to report you the bugs it found in the summary, so you can check the bug and report it correctly upstream.
Ha I reported that drag and drop bug 20 years ago. Windows does it in an even smarter way - it raises the window on mouse-down but only if the target isn't a drag source.
As I recall X11 doesn't really support that at all so there was no way to implement that on Linux. Doing it on mouse up seems sensible though. I couldn't find my bug report but I bet it was met with "I like it broken".
I've avoided that bug throughout more than 20 years of Linux usage just by disabling raise-on-click.
(Except of course the Gnome folks did they utmost to make that option unavailable back in the Gnome 2 days: Metacity's raise-on-click gconf setting was hobbled so that it could only be disabled in focus-follows-mouse mode. That bit of idiocy was patched out years ago, but the obnoxious text that accompanied it survived in Marco until earlier this year!)
> I couldn't find my bug report but I bet it was met with "I like it broken".
Which is an appropriate response when you yourself don't even want to recognize that others might prefer raising on mouse down, e.g. for lower latency in a much more common operation.
I don't have a Cameo Pro 4 to test, but I have reverse engineered software to control some of their other devices. It's a little janky but if you want to help I'd appreciate it!
I believe there is (or once was) also an Inkscape plugin somewhere but I never tried it.
As for how I reverse engineered it, I just used a USB sniffer and wireshark. I don't think there's a big risk of bricking these devices - I don't think they even have non-volatile storage, so worst case you turn it off and on again.
I would like to reverse engineer my solar power inverter too, but that is definitely brickable (the first one they shipped me came pre-bricked!) and is a lot more expensive so I've had to resist. :/
You just run `mycommand connect host` and it will connect via SSH, start a Wayland server with X support, set up the appropriate environment variables, and also automatically reconnect, so you only have to do this once.
After that you can just run GUI commands via SSH and appear locally like you always wished they would. It uses a VNC-like image data protocol so the client works on any OS. And windows are seamless - no windows within windows nonsense. I haven't put much effort into the video protocol tbh but it's already better than Xpra and very usable with 100ms latency (London to NY) so that's good enough for me.
I'll try to open source it.
I think a better solution would be if SSH itself supported some GUI channel, but the only thing is the Kitty graphics protocol which kind of sucks tbh. I did try writing some tools using that for the GUI a while back but it's awkward because you have to use PNG, you're still stuck with the character grid for text, etc.
reply