Source: Hacker News
Article note: I'm kind of hoping Wayland gets PipeWire'd (motivates infrastructure cleanup like PulseAudio did, then gets quickly leapfrogged by a more competent design once the lay of the land is clear). It would be about on schedule, it's coming up on 15 years of being the next thing.
The plan to address basic functionality like input plumbing for shortcuts and virtual inputs, pixel peeking for screen shots/sharing, screen/input grabbing for full screen programs, etc. with a flotilla of uncoordinated extensions adopted (or not) by different compositors is, was, and always has been dumb, and having it fragment among compositors instead of anchoring to a first party (or at least blessed) reference library is making it worse, so now we have a bunch of software that runs exclusively under subsets of compositors. (And I realize the pacman systemd criticism and this complaint seem contraty, but this design got most of the worst features of a single integrated system and the worst parts of an ecosystem of interoperable interchangeable parts in one go because they tight coupled the wrong things... or along the wrong axis? I'm not even sure what to call it).
MOST of the critical stuff is getting pretty hashed out, so maybe it'll settle and be the norm for a while before it collapses under the weight of questionable architecture choices. Some problems are being worked around elsewhere in the stack eg. keyd that intercepts all your HID activity at the kernel interfaces and emits events via virtual devices, which may or may not be a better allocation of trust than having the display server do it, but at least has a config format not developed by the accreted output of jibbering madmen like the XKB shit.
Comments
