I do wonder what it would be like if we have a linux distro, where you can work seamlessly between a desktop PC and a laptop or a random cafe, because the desktop is essentally a web browser like this example.
Of course you could just say IPKVM or VNC, but I would say it's a bit different if you can run most of the logic client side (via webassembly etc...). So this would require the app to be rewritten to have a clearer separation between the view and controller... which is what most linux apps don't really do. So I think at this stage it's just a pipedream.
Sandstorm really had this kind of feeling. Not that it presented as a desktop environment visually - but it offered a much more integrated “computer” of documents versus silod web site apps where you need to open each site to see the files in the app. https://sandstorm.io/
On the other side of things, you can just run an SSH + X11 server for remoting; and have fairly good integration between windows controlled by a remote process on your desktop, and windows controlled by a process on your laptop. Over fast links it’s nicer than RDP kinds of protocols.
Finally Plan 9 really runs with this idea - you can mount the processors of another system to use them on yours. Plan 9 is a distributed operating system, designed to make a network of heterogeneous and geographically separated computers function as a single system. In a typical Plan 9 installation, users work at terminals running the window system rio, and they access CPU servers which handle computation-intensive processes.
Plan B (built on top of Plan 9) and Octopus (built on top of Inferno) were closer to that - Plan 9 and Inferno had the fundamental operators but not the environment to exploit something like that, whereas Plan B/Octopus had interesting idea of "Personal Mainframe" which centralised a single person's computing experience, while making the peripherals composable.
This meant for example that you could have your "personal mainframe" in your office, then go with your "laptop terminal" to a conference hall, and when it came to your presentation instead of attaching the laptop to a display, a room-local computer could be used for its resources like screen and audio. All the while you have continuity because applications stored the state on your "mainframe"
I vaguely remember a quote from someone at Bell Labs explaining that, when they left, one of the hardest things was realising they couldn’t just turn on their home computer and just pick up what they were doing exactly where they had left it at the office anymore.
There's (theoretically) nothing standing in the way of making self-hosting of serverside apps as easy and painless as installing mobile apps, except someone has to do the infrastructure/integration work to make that UX come true. Docker got us 80% of the way there, but maintaining a complex setup can still be painful even for seasoned sysops (and we'd rather not take our work home more often than we already do). Sandstorm added another 80% and I wish it could take off like mobile did.
> Plan 9 really runs with this idea - you can mount the processors of another system to use them on yours
as i understand it, it's more that you mount your window on a cpu server so your computation-intensive process can draw into it; there isn't a filesystem interface to the processor itself
20 years ago I was using SunRays which were neat because you could tie a desktop session to a physical smart card and have it instantly pop up on any terminal you stuck the card into.
If I recall at some point there was even a SunRay laptop with wifi, but I never got a chance to use one 'cause we just didn't need it.
DonHopkins's [flagged] [dead] reply here https://news.ycombinator.com/item?id=38830583 is very interesting reading, though unsurprisingly pretty rude; if you have showdead turned off, you might want to turn it on (creating an account if necessary) in order to read it
Yes somewhat, but if I understand X11 is closer to a 'thin client' where X11 only handles drawing and moving windows on the display and interacting with a mouse, keyboard or touchscreen. This is opposite to how most a typical web browser experience works, which much of the logic happening on the client side (e.g. UI elements)
That would be a huge leap backwards. Something more like NeWS would make much more sense, since you have a scripting language in the browser that solves the terrible design flaws of X11 and Wayland.
We've come so far in the direction of NeWS by adding another layer, the web browser. Why add another third layer to move back in the opposite direction, when you could REMOVE X-Windows or Wayland and just have the browser that you need in the first place, running directly on the hardware, without the useless obsolete layer underneath.
I've written about this frequently and extensively (also extensibly), over more than 10 years on HN, and a lot longer than that elsewhere. But some people seem to have a mental block against understanding what the obvious advantages and efficiencies are, just don't get it, and never will, but actually want to move backwards towards X-Windows again.
Of course you could just say IPKVM or VNC, but I would say it's a bit different if you can run most of the logic client side (via webassembly etc...). So this would require the app to be rewritten to have a clearer separation between the view and controller... which is what most linux apps don't really do. So I think at this stage it's just a pipedream.