Post #3797492
2026-07-13 19:16 UTC
@dalias@hachyderm.io @datenwolf@chaos.social @miah@hachyderm.io What? No. It literally has the opposite effect: only the remote server (even a headless one) needs to be able to run the GL/VK APIs or accelerated encoding. It makes it so that it's tolerant of high latency connections and lower bandwidth connections.
I've literally been able to use Wayland apps with `waypipe --video ssh` over in-flight Wi-Fi over the Atlantic with minimal UX impact. The app was responsive and worked fine.
Replies (1)
-
@dalias@hachyderm.io 2026-07-13 19:28
@neal@social.gompa.me @datenwolf@chaos.social @miah@hachyderm.io You're reversing what I would call local and remote, and presuming that we want to write GL/VK stuff rather than just drawing UI elements. If you want to do that stuff like for a game engine, then yes you render it locally with a GPU in offscreen buffers and this has nothing whatsoever to do with Wayland or X; you then just submit the blit across whatever channel you have to the display. But this requires bandwidth for blitting dynamic locally rendered buffers and is NOT how normal applications should work. You are not going to make that work over 100 kbit/sec links.