GNOME Mutter Switches To High Priority KMS Thread To Avoid Crashes

[blackiwid]

You are right, that is why Qt is used everywhere, while GTK is used almost exclusively in Gnome.
Everyone is crazy in this world!

Everywhere didn't know that all windows apps are written in QT and the same for Macos and IOS and Android... why didn't I know that this is all build with Qt.

I disagree, when gnome releases in a 6 month cycle new releases they try hard to get that stable, while plasma says "ohh you tried 4.0 or 4.6 you should have waited 20 minor releases in then before obviously it's unstable, and you used the wrong distro that did not use a very very old version instead.

They use that even to critique gnome, they rather use hacks that are not fully stable / bug free to give the users features earlier than with gnome while gnome prioritizes stability, even if that means that some "powerusers" get angry about it.

And they also don't have strange uncertainties when a company that owns Qt changes their practices or licenses... Gnome clearly has more a LTS mindset than KDE.

[woddy]

Everywhere didn't know that all windows apps are written in QT and the same for Macos and IOS and Android... why didn't I know that this is all build with Qt.

I disagree, when gnome releases in a 6 month cycle new releases they try hard to get that stable, while plasma says "ohh you tried 4.0 or 4.6 you should have waited 20 minor releases in then before obviously it's unstable, and you used the wrong distro that did not use a very very old version instead.

They use that even to critique gnome, they rather use hacks that are not fully stable / bug free to give the users features earlier than with gnome while gnome prioritizes stability, even if that means that some "powerusers" get angry about it.

And they also don't have strange uncertainties when a company that owns Qt changes their practices or licenses... Gnome clearly has more a LTS mindset than KDE.

Unfortunately many Lts distributions only do bug fixes for the main DE, which is often GNOME, so with other DEs if there is a bug you keep it. However this happens a little less with GNOME, but this depends on the distribution not on Plasma.
For example, for major bugs openSUSE Leap also does bug fix Plasma but it's an exception , but this method of distributing software today really doesn't make any sense! It is not clear why bug fix versions cannot be shipped on Lts, this is a big absurdity.
Unfortunately regardless of automated tests, most bugs are discovered only after the release. Do you want it to change? Be the beta tester, because that is exactly the problem, nobody wants to test beta releases and this is the result.
The GNOME code is small compared to Plasma, it is no coincidence that most users use third-party extensions, which often have bugs or do not work at all on the new version. The extension of the icons on the desktop for example, often does not work properly or lacks essential functionality.

As for Qt vs GTK it is undeniable that the most used is Qt and GTK nobody gives a damn about it outside of Linux.

[bug77]

Well, that tells you the compositor is doing something wrong, flooding the kernel with instructions it is not able to handle in realtime. If it did throttle the commands of only sending more when previous ones finished then there would be no problems.

If would, if I had proof that's where the offending calls come from.

[bug77]

Hey, at least I didn't just compare one data point each and claimed an x.yzw...% difference, as I see done too often.

"Others are worse"... the eternal scapegoat of the mediocre.

[Espionage724]

Ok this right here is implying large-spread incompetence. GNOME switched to RT on their own, but is having another tightly-integrated component (Mutter) "sometimes" crash, because of an unrelated but larger component (kernel) being busy, and timings being involved in a manner to make this inconsistent, while every mainstream distro is shipping it? And these decisions are going through multiple chains; GNOME can do what they want but distros are supposed to be QA'ing this stuff before presenting it to users in a non-beta or even paid form.

If this explains this (GNOME sometimes failing log-in), oh boy: https://discussion.fedoraproject.org...login/118415/5

[blackiwid]

The GNOME code is small compared to Plasma, it is no coincidence that most users use third-party extensions, which often have bugs or do not work at all on the new version. The extension of the icons on the desktop for example, often does not work properly or lacks essential functionality.

So there are good reasons why it's more stable, you think that this reasons are somehow bad, because you need some from my view strange features and more complexity. For me none of the extensions if I sometimes used some of them are essential more like nice things. Currently I don't use any on my gaming pc where I use gnome.

But you can say the normal gnome is shit because for you it lacks to little feature, you can even claim that this is basically universial for 99% of people I disagree but I grant you that for this argument, that is incompatible with the statement:

But then, gtk is a well known bad basis for development.

Except you think it's just bad experience to develop for it, because stability is not the problem, which I would think first if I say "something is a bad basis for development".

As for Qt vs GTK it is undeniable that the most used is Qt and GTK nobody gives a damn about it outside of Linux.

And I don't give a damn about anything QT outside Linux, because for the smartphone I either use Android, or I recently tried postmarketos on a old ebookreader sadly runs like shit and after trying 5 different OSes the one that still totally shitty but barely somewhat usable was xfce based.

In windows if I rarely use it, I use for games only, I couldn't care less about Qt there, this is a Linux website primary so why would we care what is in some other systems very seldom sometimes used?

Except you think Windows is better than Linux because more people use it and develop for it? But why would you then try to defend Plasma because more people use windows instead?

[MrCooper]

"Others are worse"... the eternal scapegoat of the mediocre.

If you guys want to make an argument that the conclusion is wrong based on the data points in the graphs, or have other ideas for solving the issue, I'm all ears. Otherwise, taking cheap shots from your armchairs at how we arrived at the conclusion doesn't help anybody.

From where I'm standing, it's pretty clear that the data points are very similar in both cases. One might argue that worst-case lateness is slightly worse with high priority, and worst-case duration is slightly worse with real-time priority. Even so, it boils down to more or less a wash in terms of the resulting worst-case latency. I see no clear benefit for real-time priority which justifies letting users hit the compositor getting randomly killed.

If this explains this (GNOME sometimes failing log-in), oh boy: https://discussion.fedoraproject.org...login/118415/5

The issue we're discussing here only happens in a Wayland session, so doesn't sound related offhand.

The characteristic journal line for this issue is

systemd[<PID>]: [email protected]: Main process exited, code=killed, status=9/KILL​

[You-]

I love how excited the anti gnome crowd is to learn about a bug "in gnome", even if they only found out about it after it was fixed.

"Finally, I can say gnome is as buggy as what i use!"

[Topolino]

MrCooper I hope you get a chance to revisit the benchmarks when Fedora 42 gets close to end of development. Who knows what changes might come to the kernel and drivers.

[oleid]

By you and some other biased KDE Fanboys, sure.

QT is for people with ADHS, [...]

saying you've no clue about ADHS without saying you have no clue about ADHS.