Hi - reading the linked updates, I couldn’t help but wonder how recently folks may have assessed the current state of Nvidia + Wayland in Ubuntu Desktop. If 24.04 Noble Numbat will maintain the approach of defaulting to Xorg when an Nvidia card is being used, I do think it would be worth it to re-assess for future versions.
At least based on what I see in these updates, it would appear that at least some upstream (e.g. GTK, mutter) issues that are specific to Xorg are functionally deprioritized, which is understandable given where new development focus has shifted. Meanwhile - and unfortunately all I can offer is anecdotes - I have AMD+Nvidia hybrid and Nvidia-only systems that I run Wayland sessions on every day, without issue.
Those anecdotes of course don’t mean that Nvidia + Wayland should be a default for everyone, but it at least makes me wonder how such an assessment could be done in a more structured way. Is it as simple as stacking up the known issues that are Xorg-specific on one side, Nvidia-specific on the other, and determining which side’s issues are more tolerable? Are there things that are seen by the Ubuntu team as “showstoppers” (borrowing the KDE term) for defaulting to Wayland with Nvidia? Is the process for collecting and assessing those issues and proposing a decision framework something that the community could feasibly help with?
Apologies if this has already been discussed ad nauseam, but I wanted to see if there’s an opportunity to help with some structuring and presenting of the current situation and decisions to come, if it’s relevant.
There is something in that Mesa drivers should not be the same for LTS like hwe default Mesa version with default (nvidia) drivers but also Intel/AMD. But in linux world there is a lot of other things involved but some driver update stack is required.
One of the solution is snap mesa stack but also flathub has its own version of drivers(but try GNOME builder and build something and for a few times and 10GBs of not needed libs are inluded on your drive :). Thats something I don’t prefer and if some stack is hwe included that could benefit not only me but maybe also someone like you
Once explicit sync makes its way into 24.10 then I think defaulting to Wayland would be a good idea. All of my problems with NVIDIA & Wayland have resulted from lack of explicit sync.
Basic things like copy paste between some apps (Git Cola → CodeLite for example) still do not work on Wayland. Screen sharing during video calls (Slack) still does not work.
Why even talk about making it default on any system when it’s still broken for many normal use cases?
Because a) it’s a default, which can be changed. b) being default, helps put pressure on software vendors c) it’s not an LTS release. So issues affect a much smaller audience.
I get what you mean, but that only applies to technical people. Try to think about a normal user who is installing Ubuntu for the first time, only to see that they can’t share their screen.
They don’t know what “X11” or “Wayland” is, these are technical details. All they know is that their apps are broken, and it didn’t use to be broken.
There should never be a public release that knowingly introduces broken items that were not present beforehand.
The pressure should not be coming from single users → software vendor, it should be coming from Canonical & RHEL in the form of paid work to Slack Technologies, etc to get it to work on Wayland before switching.
Fair enough to have your opinion, but I do believe that you might be underselling the salience of popey’s point that it’s not an LTS. The non-technical folks you describe should - and likely already are - using LTS releases only:
An estimated 95% of all Ubuntu installations are LTS releases.
So I’d argue the interim releases, much like Fedora’s releases, are in fact perfectly positioned to create some public pressure by letting such developers know what’s coming down the pipeline for wider adoption.
(Setting aside the philosophical question of whether Canonical, Red Hat, etc. should pay for an individual app developer to bring their app into standards compliance, which I would disagree with just based on the precedent it sets)
Sorry @johnandmegh for missing your original point and being too terse before. I have been in a rush a travelling for most of the past month…
The list of Nvidia-using-Wayland-issues has been reviewed and revised repeatedly over a few years now. It might seem like an overly simplistic approach but simplicity is a good thing. And the list of issues used to be much longer than it is now. You can’t see the progress now, but it has progressed.
Help is always welcome, and anyone can contribute to the ongoing assessment simply by adding the tag nvidia-wayland to a Launchpad bug. That tags means “this bug only occurs on Nvidia, and only in a Wayland session”.
Minor tangent in response to the later comments:
While non-technical users “should - and likely already are - using LTS releases only”, some people are also sufficiently non-technical that they don’t know the difference between LTS and non-LTS and so will choose the highest version number.
Taking this a step further, I find a proportion of non-technical users also don’t look at version numbers at all. And might for example install 21.04 after 22.04 already existed. I feel this particular case might be a matter of unfortunate branding where people fail to notice or assume there’s any significant difference between those numbers.
So the assumption that only technical users are on non-LTS releases doesn’t hold up in my experience. But I believe the usage stats still support the fact that most people use LTS.
No apology needed at all, you took personal time to efficiently answer a question from someone who is merely a curious user! If anything I’m at fault for assuming that such a list would necessarily have to be manually compiled, rather than just being systematically tracked via Launchpad. And it looks like only 1-2 bugs remain that are at least high importance, which to your point looks really good for 24.10!
And that’s very useful context about version number awareness - I keep forgetting that there’s a middle ground of folks who are just technical enough to follow directions and get “something” installed, but not enough to understand the factors that should go into deciding “what” to install.