shitface posted:

Tankakern posted:

gentoo just got a profile upgrade

https://www.gentoo.org/support/news-items/2024-03-22-new-23-profiles.html

bah, it wants you to do a world rebuild

Athas posted:

I am pretty sure it was originally named Phoenix. I remember installing it under that name when I was a teenager.

Well Played Mauer posted:

Thanks for subscribing to FIREFOX FACTS

DID YOU KNOW

Firefox was originally named Firebird, but had to change its name due to a lawsuit they settled with Pontiac.

Last Chance posted:

unsubscribe

BlankSystemDaemon posted:

unsubscribe firefox-facts

Well Played Mauer posted:

Thank you for updating your subscription preferences. Please allow up to 10 business days for us to completely remove you from our system.


DID YOU KNOW
Mozilla, developers of Firefox, maintain a shelter for abused foxes on their business campus in Lincoln, Nebraska. They only upgrade a full version number when a new fox pup is born. This is known as vulpine versioning and is commonly used in open source software projects.

Antigravitas posted:

You know what I want when looking at my bookmarked threads? A Reply All button.

Joshua-Ashton posted:

Revert "video: Prefer Wayland over X11 (take 2!)"

Description

This reverts commit f9f7db4

Wayland has a myriad of unresolved problems regarding surface suspension blocking presentation and the FIFO (vsync) implementation being fundamentally broken leading to reduced GPU-bound performance.

That is not to say "we should fix FIFO in Mesa/other drivers," but rather that it is completely unfixable without an additional protocol, in this case fifo-v11.

Without this protocol, vkQueuePresent or glSwapBuffers must stall for the 'frame' callback after presenting an image. The only reason we can get away with this on SteamOS is because Gamescope implements what is essentially fifo-v1 and we use that there.

The other side is surface suspension -- a very similar issue to the above wrt the frame callback being used in that way and blocking. If the SDL window is obscured, vkQueuePresent will block in FIFO, which games typically do NOT like. This is solved by the combination of fifo-v1 and commit-timing-v1.

There is no advantage to games and average applications preferring Wayland over X11 -- only severe performance and unusability regressions right now.
Thus, we must revert this change until fifo-v1 and commit-timing-v1 are released and at least in a stable release for major compositors.

Truga posted:

i've been gaming on wayland for over a year now and it works fine because games/proton still all run through xwayland anyway

i also really like that wayland kwin throttles apps on inactive desktops, means that if i i have to do something while playing, it makes the machine go basically idle. but apparently this is bad?

Cybernetic Vermin posted:

defaulting to xwayland to keep things working is indeed the point here.

and, yeah, just blocking the rendering thread is indeed bad, primarily because it doesn't work for tons of poo poo (i.e. a lot of games just crashing or otherwise getting screwed up, since that's not how things behave anywhere else), but it is also kind of dumb design that compositors take on power saving policy, and implement it by entirely blocking a thread in an application at an undocumented point (indeed for vulkan it is documented that this *must* *not* happen). it is the schedulers job to decide about that stuff, and no scheduler has historically worked like "oh, you did x, but i don't think x is meaningful at this point, so you'll just never run again now, no matter what you intended to do after the nop of x".

Truga posted:

so what you're saying is, kwin should have full access to the scheduler instead?


idk if that's a great idea, they can't even make games run right lol

Beeftweeter posted:

the reasoning might not be the same but i think this is what microsoft ended up doing with aero. i remember testing longhorn (vista) on a not terribly old at the time pentium m laptop with a GMA 900, and after installing a new build aero suddenly wasn't supported. turned out that was because the GMA 900 doesn't have a scheduler but the GMA 950 does

e: of course they reversed course a few years later and now you can have aero without any gpu at all but that solution is very very similar to LLVMpipe, so i guess i can't really complain. on that note they should try and shoot for what MS does with that (WARP10+) and have it render whatever the actual gpu can't

Truga posted:

i mean, i'd love it if window managers could just suspend or highly limit cpu cycles to apps that aren't on the desktop right now, but you know it'd be a massive nightmare for years after such a change because poo poo like pipewire also isn't on the desktop while being in the desktop session process tree

seeing how gpu usage is like 80% of the load on your normal desktop pc, just preventing something from rendering as much was probably the easiest and fastest solution lol

Cybernetic Vermin posted:

i think you're thinking of the introduction of the os gpu scheduler, which indeed happened with vista (wddm 1.0),

Cybernetic Vermin posted:

windows *also* does scheduling based on which windows are active, has since 95 i think, but i think the advantages there are modest (done more for interactivity than for energy saving).



more generally obviously pausing and then tombstoning background applications is indeed how you'd do anything new today, but you then need a proper infrastructure and api for application lifecycles (akin to ios or android), and you're not going to make existing programs play nice with that without great effort. waylands approach on that is a really crude middle ground which fails to make sense both as an application lifecycle thing and as a "classic desktop" thing. i expect it'll get fixed though, and indeed workarounds exist. it is annoying how long what i view as severe blockers take to work their way through though.

Tankakern posted:

a little while until wayland is there for gaming then

Antigravitas posted:

The compositor giving scheduling hints seems to me like a good idea actually. The compositor knows which applications are currently being shown or even active, and those should absolutely be given priority over other tasks.

The_Franz posted:

they can easily suspend themselves if they want, now they just need to stop forcing it on them.

Truga posted:

i mean, i'd love it if window managers could just suspend or highly limit cpu cycles to apps that aren't on the desktop right now, but you know it'd be a massive nightmare for years after such a change because poo poo like pipewire also isn't on the desktop while being in the desktop session process tree

seeing how gpu usage is like 80% of the load on your normal desktop pc, just preventing something from rendering as much was probably the easiest and fastest solution lol

Beeftweeter posted:

so is job control just not an option here because of the way the process group is split into scopes, or for some other reason? sending something SIGSTOP should just pause it until it gets SIGCONT... right??

or am i just babbling here

Truga posted:

games already release 30%+ unfinished, nobody's gonna code a frame limiter for when you alttab lmao

sb hermit posted:

what if it’s a networked application or a music player or something else that requires background activity?

what if someone or something else sends SIGCONT or SIGSTOP while it’s running? Should the compositor ignore previous state? And how would the user know enough to call SIGSTOP and SIGCONT on the application to get it to behave correctly again?

I think a lot of applications would rather block than receive signals. And I think SIGSTOP and SIGCONT should probably be used judiciously rather than a routine action by a service.

Truga posted:

that means that every app has to do power saving by itself, which is *never* happening lol

Beeftweeter posted:

modern windows apps can definitely manage power saving, but i'm not sure how they do it. i imagine most of them do it manually since they might be running on (relatively) old windows

but for example every time i try edge on windows it tells me to enable its "green mode" or whatever, and it is somewhat granular in what it does to achieve power savings. but since enabling that fully also has edge start with the computer as a background process, lol no thanks