Hacker News Reader

Comment by cturner | Hacker News Reader

Remote windowing is a legacy of the era when universities ran dozens of dumb terminals from a server. It would be bizarre to pursue that in an era of hardware-accelerated graphics.

If the Wayland design was weighed down by such things, it could never hope to be progress.

View on HN

Universities still do that. It's not that bizarre of a use case

It is actually a common enough use case that Windows supports it now via Remote App, which is bit of a hack over RDP since RDP wasn't really designed for that. The use case may be relatively small but it undeniably still exists.

It is reasonable to want remote desktop access. I am unfamiliar with the RDP technique you raise, but expect it could be achieved through VNC or similar systems.

X-Windows is itself a weak response to remote access. I have memories of trying to use Oracle Forms over a modem connection to uni, and finding it far inferior to VNC connections to work systems - effectively unusable.

The remote-windowing mechanisms of X-Windows are coupled to aspects of X-Windows that Wayland was created to overcome. The video 'The Real Story Behind Wayland and X - Daniel Stone' discusses this.

Microsoft Windows does not do remote-windowing in the X-Windows sense. RDP is not Explorer. From what I read, a RDP-like-thing could be built, and Wayland would not get in the way of this. I think the Wayland team make a clear case for why this is out-of-scope for them.

this is sort of exactly my point but inverse - X windows is awesome on a lan & due to the 'native' support for individual applications, but there is definately huge room for improvement w/r/t display protocol performance / latency issues which are largely a result of relying on X server drawing architecture

don't throw the baby out with the bathwater, in other words

But this is the thing, X is not network transparent anymore. Basically no app created in the last decade use the native rendering libs of x, and thus what happens over the network (moving bitmaps) is much worse than what a decent video streaming algorithm could do.

This ignores the current tend of hosted gaming services. The pendulum regularly swings back and forth between thin and fat clients as cost and innovation evolve.

There is no reason for a display protocol to know about these things. At most it is responsible for giving access to a window’s/screen’s buffer and then any app can do whatever they want with it, eg compress it and stream it (which as I wrote in an other topic is usually better performing than today’s frameworks over whatever remained of X remoting, with bitmap transfers and the like)

Quite true. I've have to switch my desktop back to Windows because the only remote gaming server worth using (Parsec) doesn't work on Linux.

How many people use TeamViewer, Anydesk, etc? I don't see how remote windowing is legacy. The feature would be a huge hit if it were modernized and made more accessible and efficient (akin to Microsoft's RDP), and that's only possible if it's a core part of the protocol.

TeamViewer/AnyDesk/VNC is not "remote windowing", it's "remote access" to a whole desktop. And that's easily available for the wlroots ecosystem https://github.com/any1/wayvnc

But actually "remote windowing/apps" is even better supported, it's a universal proxy: https://gitlab.freedesktop.org/mstoeckl/waypipe Absolutely does not require any support from the core protocol.

Reply on news.ycombinator.com