GNOME Shell + Mutter See Big Last Minute Improvements With The GNOME 3.36 Beta

[sarmad]

"- Adding a new Blur shader to the Shell."

What is this supposed to do? Is it about adding blur effects?

[brent]

I really hope they land a lot more stuff! The most important area of GNOME, performance, still didn't get the attention it deserves. And this time it's not just Daniel's MRs that have been stalled. E.g. fullscreen unredirection on Wayland has been quite ready for months, but doesn't get any attention from maintainers. Many of the performance-oriented MRs would only require a little bit of additional work, if any. It's a bit sad.

The gnome-with-patches copr shows the potential of the current set of performance patches: with the patches, GNOME runs at solid 60 fps at all times, for the first time. Doesn't matter if you have 100 windows in the overview or if a heavy build runs in the background, it's still nearly flawless.

[brent]

"- Adding a new Blur shader to the Shell."

What is this supposed to do? Is it about adding blur effects?

There's a new lockscreen design in development that uses a blur effect. The current design looks really dated, so it can only improve.

[cynical]

But I don't know. Maybe I'm just old, and getting to programs simply and quickly doesn't matter anymore. And I'm not being facetious. From what I've seen many people just place shortcuts somewhere, and if they don't have a plethora of them it works okay.

Or maybe people simply don't agree with you on what is efficient and simple. Pecking through menus is the true abomination. Either you run the app frequently enough to put it on your bar, or you will find it in one click in the recent apps screen. If you already know what you're looking for, you can hit the meta key, type two or three letters, and then hit enter to launch it. It could not be easier. You can even guess the app by typing "file" if you want the file manager, or "music" for the music player.

You are just used to the old way and don't want to change your habit. That's fine, there are other environments you can use.

[TemplarGR]

Windows 8 and Windows 10 introduced what I call "The Age of Convolution."

This is the time when simple and direct point and click interfaces, where GUI efficiency was easily measured by the average number of scrolls and clicks it took to access any program, was replaced by who-knows-what, and the subsequent requirement that users type in hoped for commands in a search box like good old DOS, or look it up on the Internet, to find out where it is.

And sadly Gnome was one of the the first in Linux to follow this confused paradigm, though Unity was the first, in, what I consider, to be Linux abominations to GUIs.

But I don't know. Maybe I'm just old, and getting to programs simply and quickly doesn't matter anymore. And I'm not being facetious. From what I've seen many people just place shortcuts somewhere, and if they don't have a plethora of them it works okay.

But, again, I don't know. Between Windows 10 and most Linux desktops today, it seems that while the Age of Convolution predominates, it is worse, not better.

Actually, this paradigm is one of GNOME's selling points and strengths. Hitting super and pressing 2-3 keys to find any app you want and open it afterwards with a single click is far more efficient than any old style menu/submenu hierarchy. Any often used app can be pinned on a bar and for the rest you just need to vaguely remember its name or its function.

The main issue with GNOME is that it is not efficient if you need to work with many windows and switch focus often between them. It needs a proper autohidden taskbar in my opinion.

[Teggs]

Hitting super and pressing 2-3 keys to find any app you want and open it afterwards with a single click is far more efficient than any old style menu/submenu hierarchy.

Nice strawman. Muncrief wasn't speaking about using the menu system, he was speaking about double-clicking a desktop icon.

I have never seen anyone argue that the ability to search for a program by typing its name was bad or should be removed. I repeatedly see people saying that desktop icons are bad and should be removed. These arguments often include that holding down a key and pressing three others, or opening a search bar and entering characters, is faster than moving a mouse and double-clicking on a screen object: an argument that was false, is false, and will remain false in future.

I wonder what is really driving this push to kill desktop icons.

[wizard69]

Windows 8 and Windows 10 introduced what I call "The Age of Convolution."

This is the time when simple and direct point and click interfaces, where GUI efficiency was easily measured by the average number of scrolls and clicks it took to access any program, was replaced by who-knows-what, and the subsequent requirement that users type in hoped for commands in a search box like good old DOS, or look it up on the Internet, to find out where it is.

And sadly Gnome was one of the the first in Linux to follow this confused paradigm, though Unity was the first, in, what I consider, to be Linux abominations to GUIs.

But I don't know. Maybe I'm just old, and getting to programs simply and quickly doesn't matter anymore. And I'm not being facetious. From what I've seen many people just place shortcuts somewhere, and if they don't have a plethora of them it works okay.

But, again, I don't know. Between Windows 10 and most Linux desktops today, it seems that while the Age of Convolution predominates, it is worse, not better.

Gnome could use some work but why make it carbon copy of every other GUI/DT? Especially on Linux when you can have several different DT environments available and use what is your liking? If I get tired of Gnome there are others I can boot to at log on time, it isn't like I'm stuck with anyone interface.

[Isedonde]

brent Asking for not-ready-yet MRs to get prematurely merged is just another example of instant gratification.

I agree, it would be a very bad idea to merge MRs which are not yet ready. I definitely support that decision.

However, I still feel like mutter lacks review capacity / maintainers / mergers. Take this MR for example: https://gitlab.gnome.org/GNOME/mutte...e_requests/724
It was proposed 5 months ago, so well before the previous 3.34 release. As is to be expected for such a change, there was some initial discussion and clarifications of the change, which is a good thing.

Then, 3 months ago, there was a reply "we want to merge this, but not now, because someone is working on something related and he'd have to drop parts of your new code, and we want to land that change first and then this will be merged later in 3.36". Now as is very common in software development, that other change is apparently a bit late (yay deadlines). This is not a problem because we all know making software is hard! But it means that the finished work by van Vugt is still waiting since longer than the beginning of the cycle to be merged. Maybe priorities have shifted and the other related change is not expected before 3.36. Should that 5 months ready MR really be pushed to 3.38 or later even though it was ready even before the 3.36 branch started and is an important performance improvement?

Also 3 months ago, van Vugt noticed an animation performance issue and investigated if it might be related to this MR. But it turned out that it's just a gnome-shell slowdown in other parts of the shell.

Other than that, this MR has just been sitting and waiting, with van Vugt repeatedly having to update his code to match current git changes.

Now this MR: https://gitlab.gnome.org/GNOME/mutte...e_requests/983
It's not as old, but still: December of last year. Everyone agreed that the change has positive effect. There was some discussion about making the change in pango, but apparently pango maintainers have not come around to do anything about it in that time yet so if the goal is to improve 3.36, the "fix in pango" way is probably not possible. A bug was found 2 weeks ago and fixed 2 weeks ago. Now it's silence with no comments by maintainers like "this looks good, merging" or "this is good for performance, but from a maintenance POV it has to go into pango, closing". It just sits there and waits.

Obviously I'm not saying that current maintainers are lazy / should work more. I'm sure they are busy with other things. However, I do think that it would be good to improve available review/test capacity for important projects like mutter, either by having more maintainers or by reducing the amount of other responsibilities the current maintainers have. It's a bad sign when "ready to merge" MRs rot in the MR tracker without any positive / negative comments by maintainers for weeks on end. I mean, when I look at other "many Linux desktop users use this quite important bit of software" projects like Mesa, MRs are quite often reviewed and merged within a few days or even hours.

[TemplarGR]

Nice strawman. Muncrief wasn't speaking about using the menu system, he was speaking about double-clicking a desktop icon.

I have never seen anyone argue that the ability to search for a program by typing its name was bad or should be removed. I repeatedly see people saying that desktop icons are bad and should be removed. These arguments often include that holding down a key and pressing three others, or opening a search bar and entering characters, is faster than moving a mouse and double-clicking on a screen object: an argument that was false, is false, and will remain false in future.

I wonder what is really driving this push to kill desktop icons.

LOL, talking about "strawmans" while strawmaning himself. Dude, he was NOT speaking about double-clicking desktop icons. His rant began with Windows 8/10, which SURPRISE have the same double clicking desktop icons since the earliest versions of windows, nothing changed in that regard.... And it is obvious from his rant that his main issue is the start menu or lack there of.

By the way, desktop icons are garbage. There is no reason to have them at all, i don't use them on Windows, i don't use them on KDE, so i am not missing them on GNOME. They are just a clutch for lazy people.

[curfew]

Nice strawman. Muncrief wasn't speaking about using the menu system, he was speaking about double-clicking a desktop icon.

I have never seen anyone argue that the ability to search for a program by typing its name was bad or should be removed. I repeatedly see people saying that desktop icons are bad and should be removed. These arguments often include that holding down a key and pressing three others, or opening a search bar and entering characters, is faster than moving a mouse and double-clicking on a screen object: an argument that was false, is false, and will remain false in future.

I wonder what is really driving this push to kill desktop icons.

You shouldn't need to lie in order to defend your position. Desktop icons are most often covered by many windows, so most users will likely minimize each obstructing window one by one to uncover their desktop. More advanced users might hit a button on the taskbar (if available) or have memorized the keyboard shortcut for it. Oh, did I just mention typing on the keyboard as a requirement for launching an app? ::facepalm