Harshness towards gnome and gtk

Hi to all

I read this article:
https://forums.theregister.com/forum/all/2022/07/05/gtk_5_might_drop_x11/

And the comments.

The comments are really harsh and I would like to know if this is the thought of majority or maybe the readers of ‘the register’ are for some reason all gnome detractors…

One of the arguments are the compatibility breaks
Another one is the lost touch on user’s needs
And so on

I love gnome/gtk/libadwata. And I think that it’s superior to any other toolkit

Just my two cents

The Register tends to have of people who prefer the older, less modern way of computing.

They were the same people who were very critical of Rust in the mid 2010s, stating that C and C++ exist so Rust will never get big. They are positive about Rust now though :sweat_smile:.

When you consider what GNOME is and does, it’s not really a surprise that The Register’s audience doesn’t like GNOME and the decisions GNOME make.

It is what it is.

All that matters is whether you enjoy using GNOME or not :smile:.

1 Like

The question is, compatibility with what, or for what?

I agree with you
And as I wrote I like very much gnome/gtk and used it and developed applications with it since day one
I really hope that it will develop more and more and in different environments, like for example embedded devices.

With previous versions I guess

I read over all the responses. While it’s surprising to me that they even think about it, so it would be indeed very sad and for my kind of Unix style apps and work the end of using it and keep porting my major projects to latest Gnome/gtk which I learned and love since gtk1! Unless – which I ain’t know much about at all but enough so I do and will and can not use Wayland – features compatible with X11 display export (ssh -X …) and some kind of X11 window manager to have control over app windows and positions will get available on Wayland natively. Which I currently do not see or heard about.

I won’t mind a all new and improved “ssh -X” mechanism. Or next gen all new X11 or Wayland. But please on a app basis. And super sweet would be a way to mirror or move running apps transparently between local and remote displays !! Without video streaming the whole desktop what has its limitations and problems with major resolution or display size miss matches. Or even head less remote machines…

And to put more oil into the fire…
I am more and more annoyed by the simple and effective old X11 style mark and paste anywhere via middle mouse button going away or been not support by more and more newer apps or what ever is behind that.

And did I mentioned: gtk4 (I know exactly yes) is already horrible ineffective and 10x slower than gtk3 when running in forced cairo emulation mode when exported via ssh -X. So here we go… already now. -X support not been given any love any more.

This is a fact. I measured it. Varies by exact widget use and type. Pop-ups are particularly poor performing.

Are we targeting now more “MS clients” aka the Word and Exel using world and secretary’s letter writing need, gamers, … vs. a powerful universal research and more tool?
Used to tackle Mars missions and possibly control Webb… :frowning_face:

Consider having a look at waypipe.

3 Likes

Looks promising, I need to check that out and see if usable for work… thanks for the link.

@jfrancis this indeed does look very good! I build waypipe and installed it on both machines not even running wayland at this time.
I was just able to open my “production level” gtk3 application on my microscope server using waypipe within a plain weston window (on my office X11/Debian/Gnome machine) for a test on my local network. So far so good. Next step is to check how smooth it goes via long distance LAN via my external provider (400/50 MBits).

Just a few more notes. On X11, middle mouse to paste is also an optional convention, and clients do not have to support it. The Cairo renderer in GTK4 is considered a fallback mode and it has degraded functionality on every backend. The main renderer for GTK4 is the OpenGL renderer.

A few thoughts about cut and paste:

The middle mouse paste is a lovely feature – I wish I’d not need to do anything to have it for any app, means text widgets, etc. should just support it. – Or in a more generic and clean way “cut/paste events” or actions ideally would be a system settings option to select what behaviors or keys you want for that. Crtl-C/-V, mark and MB or both. So every one can pick what he is used to or try. I bet so many new to Linux do not even know about it.

Not at all “individually” as this leads to chaos when every app doing eventually it’s own thing.

I’m not going to address the “ssh -X is unsupported” bits, except to say that X is a terrible protocol for drawing anything that isn’t a research tool written in 1995 using Motif, and wrapping the X protocol over an SSH tunnel isn’t going to magically make it any better. You’re much, much better off using remoting protocols like RDP. Even VNC is a better choice than remoting X calls with any modern toolkit—and by “modern” I mean “written in the past 20 years”.

What I’m going to address is this little question:

Now: if I thought that getting rid of the X11 backend gave us the ability to reach the vast majority of users on this planet at the expense of the minority of university students and researchers, even at NASA, I would do that in a heartbeat. It’s not even a question worth pondering. Targeting a niche of a niche of a niche use case is only worth it if the occupiers of that niche are also willing to put their money where their mouth is, so to speak; in other words: if people who need niche support are also willing to work on it.

In any practical sense, though, the question is a false dichotomy: we can (and should) support the vast majority of use cases, since that’s where the users are; and “using SSH tunnelling of X11” is not a use case, niche or not. It’s just a class of users who haven’t adapted to the changes in the technological landscape. There are perfectly functioning replacements for that use case, but they require changes, and typically those changes are easy to resist until they aren’t any more, at which point come the recriminations.

I am aware the -X (x11 protocol) been old and not most efficient any more with new mainly bitmap or even GL oriented (what does not work at all via the remote (-X) X11 protocol) applications. But it’s plus is it simply works.

The key issue why I am still using -X are the problems arising when the remote desktop is huge (say 3x 4k screens) and the client is a not 1:1 match in resolution (or superior) – I have not yet found a good solution for. I am not taking nor need to export the whole desktop. Even at times that would be useful option. (Like a hardware based “KVM via network link”).

Also exporting or remote using a small single or few windows only app is generally easy via vnc or various efficient tools for a simple single display.

I need to open multiple applications and one with many control windows required side by side in any layout suitable for the task. And those may span 2…3 4k displays. I had not good luck trying VNC . Mostly it totally garbled my display / Server configuration… simply not user friendly. Nor do I want to install many 3rd party more or less stable tools. Example desktop screenshot with multi window Gnome/Gtk application:

The good old -X is simply there and a native part of X11 and works – one big plus.

And with many windows it does a OK job, even not the fastest.

This said, if there is a bright solid future with Wayland and Waypipe or things a like may be getting main stream – I am all in for it! Don’t get me wrong – I am all for progress and innovations! Only unhappy when mission critical features get eliminated before stable new+better alternatives are available.

Going back to the original question…

So, here’s the thing on that:

  1. An article like that is practically inviting negative reactions. It even says up front that the move is bound to be controversial, and all but sets the stage for a less-than-enthusiastic conversation in the comments.

  2. At the same time, the change itself is only going to get two types of reactions: Negativity and indifference. Think about it. From a user perspective, why would anyone be positive about the decision to drop X11 support from Gtk 5? It’s a change that purely helps developers. There ARE NO positive aspects to the change, from an end-user perspective. So the reaction is going to be split into two camps:

    1. People who don’t care at all (who aren’t going to comment), because it doesn’t affect them.
    2. People who react negatively to the change, whether it actually affects them or not.

So, honestly, if you look at it from a user perspective, the negative reaction should be EXPECTED. I’m not saying that’s a reason not to make the change, and I’m not saying that it means the change isn’t the right move. But anyone who’s surprised by the backlash (even more unbelievable considering El Reg very clearly predicted it right in the article)… well, that would really show they’re out of touch with the users.

Well, sure, because it is a compatibility break. Sometimes dropping legacy support is OK, and sometimes it’s the right decision, but it always means that some users get left behind, or are forced to catch up against their will. And any change like that always meets some resistance.

Which may or may not be fair. It’s partly a question of numbers: How many people are still running X11, vs. how many are on Wayland? For the ones still running X11, realistically, how many can migrate to Wayland without being negatively impacted, vs. how many will be inconvenienced by the move?

(I switched from X11 to Wayland on my primary desktop only with the move to Fedora 36. I run an Nvidia GPU, so moving from the proprietary drivers to Nouveau did mean sacrificing a number of features. Fortunately, my GPU is relatively underpowered, so I didn’t lose much. But if I was in the habit of doing a lot of hardware-accelerated video encoding, or rendering, or had a card with CUDA capabilities I relied on, then the switch would’ve severely disadvantaged me.)

Nvidia is working (finally) to get their proprietary drivers working with Wayland. From what I hear, it’s going… “eh”. Assuming they eventually succeed in squashing the bugs and making it a seamless transition, then that’s a major class of users who can leave X11 behind without any significant loss of functionality. And that would make dropping X11 support much less controversial. But talking about it now, before that happens, will of course be seen as premature by many people. Whether or not it actually is.

…From what perspective?

As an enduser, I have to say it’s supremely weird that you would really care one way or the other. What toolkit an application is written with could not possibly matter less to most users, because why would it? They just want software to work for them.

Ideally endusers wouldn’t even know what their application is built with under the covers — in fact, I would argue that, to whatever extent they DO know, it represents a failure of those toolkits to properly do their job by staying out of the user’s way. The only time a user ever has to care that some piece of software is written in Gtk or Qt or wxWidgets or what have you is when that fact causes the application to fail them in some way. If everything is working as it should, the user never notices, and shouldn’t notice.

From a developer perspective, well, you’re certainly entitled to have preferences when it comes to development tools, environments, frameworks, etc. And you should definitely use whatever you feel best serves your needs and is best suited to the needs of whatever code you’re writing. But don’t lose sight of the users’ perspective, including the fact that they will never applaud you for choosing to develop software using Gtk, because it shouldn’t MATTER to them. And if we all play our parts right, it won’t.

3 Likes

The end users would have assurance that no GTK5 app is using X11 APIs anymore, so none of them could potentially cause a security issue with X11. Those apps would also be able to remove the insecure --socket=fallback-x11 flatpak option.

But IMO the reactions are unnecessary, if it really came down it, I am sure someone would figure out how to do something like using Weston to run the apps again on X11. So this would never be a thing that the end users have to worry about, either. Users who opt into that know they are running an insecure setup and can accept the risks.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.