To maybe prevent a catastrophe: The system is not able to restore virtual desktop assignments yet, it only starts the apps you had open before.
There have been incidents involving malicious code downloaded through Plasma global themes.
No malicious code was involved, just buggy code.
the fact that 1.8 was working tells me that it is possible for a window manager to work well for nvidia
Nope, it's a race condition for which the visible effects can appear or disappear for plenty of reasons. The only fix is explicit sync, which is being worked on for wlroots
Most displays provide settings to modify the colors of your screen; mine has like 10 different "picture modes" that strongly modify gamma curves, colors and the whitepoint. The EDID only describes colors of one of them, so if you change display settings, the data no longer applies.
More generally, the information isn't used by Windows or other popular video sources by default, so manufacturers don't have much of an incentive to put correct information in there. If it doesn't make a difference for the user, why would they care? Some displays even go so far as to intentionally report wrong physical size information, to make Windows select the default scale the manufacturer wants to have on that display (or at least that's what I think is the case with my cheap AliExpress portable monitor)...
That's not to say that the information is actually often completely wrong or unusable, but if one in tenthousand displays gets really messed up colors because we toggle this setting on by default, it's not worth it. We might add some heuristics for detecting at least usable color information and change this decision at some point though
Widgets aren't themes. They're things on your desktop that people are using for example for showing a folder - and if that can't interact with the system, that widget's functionality is broken.
Of course, that should not apply to install scripts or the like, which shouldn't be a thing at all really. And it should be made a lot more obvious which downloadable things can execute code / which ones are "guaranteed" safe and which ones may not be.
FreeBSD isn't working on a Wayland port, that's already happened. The Plasma Wayland session has supported it for quite a while... KDE even runs a CI job on FreeBSD for every merge request, where kwin_wayland autotests are run.
Considering the amount of complaints we got when something broke recently though (which is to say, none), it doesn't look like it has a lot of users
Thst might change with Flathub's ambitions to become an actual app store though
Neither of these things are true, if you're using Wayland for both sddm and the session
No. Source 2 still uses some X11 specific stuff and has to be ported over to Wayland before it can work Wayland-native
Another issue I have is that my touchpad doesn’t get detected if I’m holding down a key
That's a libinput feature, meant to prevent you from accidentally using the touchpad when you're typing. You can disable it if you want.
Why collect telemetry at all if you're not going to use it anyways?
Because we're sadly not collecting enough and actually useful telemetry. I think we know from telemetry that a big majority has double click set, but we don't know why (default setting vs user chose it explicitly).
And we can't easily add such things without breaking user trust. We need a new telemetry system that's more useful and extendable, but doing that is a lot of effort that noone has put in yet.
And if the distros are already having it as a default, it's even worse. We're setting it per default because it's set as the default.
No, we're setting it as a default because a bunch of people that are closer to the users than us decided to deviate from the upstream default. That's a super clear sign that we're doing it wrong
Zamundaaa
joined 1 year ago
That's a terrible example, because they completely ignore the many many years old standardized APIs (screen casting and remote desktop portals) that they could use, in favor of doing hacky and broken things that require root access instead.