Created attachment 176239 [details]
Screenshot 2 (Plasma 6.0.5)

Created attachment 176240 [details]
Screenshot 3 (Plasma 6.2.4)

Created attachment 176241 [details]
Screenshot 4 (Plasma 6.2.4)

Hi - a few questions just to check and clarify:

Are both machines running postmarketOS? Do you have access to any other machines with other Linux distributions to check against? (I can't reproduce cursor sizes that mismatch what's set in System Settings on my machine, Fedora Linux 41, hence why I'm checking)

Could you please provide in your Steps to Reproduce the cursor sizes that you're setting in System Settings?

Do these issues persist if you use the default Breeze theming everywhere?

Thanks!

Created attachment 176431 [details]
Screenshot showing the exact "Classic Flat White" cursor theme

I'm using the same cursor set on both machines, "Classic Flat White", I'm hereby adding a screenshot which exact one it is.

The machine in Screenshot 1 & 2 uses a DPI display scale of 175%, the machine in Screenshot 3 & 4 uses a DPI display scale of 150%. However, since DPI scaling is meant to scale everything proportionally, I wouldn't expect the cursor to be suddenly be harder to see because I picked the "wrong" DPI scale, so if this is in part the cause this feels like a bug.

When I pick the "Breeze" default cursor, Screenshot 4 with Firefox on Plasma 6.2.4 is semi-fixed in that it now shows the same cursor size as in Screenshot 3 with Plasma task bar on Plasma 6.2.4, but that still seems to small compared to Screenshot 1 on the other machine which feels like the correct size. Generally, with Plasma 6.0.5, picking "Breeze" doesn't seem to change anything.

Here's a table:

- Plasma 6.0.5 "Breeze" cursor 175% DPI display scale on Machine 1 on task bar: what feels like correct size 
- Plasma 6.0.5 "Breeze" cursor 175% DPI display scale on Machine 1 on firefox: ludicrously tiny size
- Plasma 6.2.4 "Breeze" cursor 150% DPI display scale on Machine 2 on task bar: what feels like ~20% too small
- Plasma 6.2.4 "Breeze" cursor 150% DPI display scale on Machine 2 on firefox: what feels like ~20% too small
- Plasma 6.0.5 "Classic Flat White" cursor 175% DPI display scale on Machine 1 on task bar (=Screenshot 1): what feels like correct size 
- Plasma 6.0.5 "Classic Flat White" cursor 175% DPI display scale on Machine 1 on firefox (=Screenshot 2): ludicrously tiny size
- Plasma 6.2.4 "Classic Flat White" cursor 150% DPI display scale on Machine 2 on task bar (=Screenshot 3): what feels like ~20% too small
- Plasma 6.2.4 "Classic Flat White" cursor 150% DPI display scale on Machine 2 on firefox (=Screenshot 4): ludicrously tiny size

I currently don't have another desktop distribution install available for testing. I hope I didn't accidentally miss a question, let me know if this helps.

I forgot to mention the cursor size, "Classic Flat White" doesn't allow setting any different sizes. I don't think I ever changed the size of cursors on any plasma install.

Small cursors appear to be a general issue with the Classic Flat White theme - ex. the most recent review of the one without shadow (but otherwise presumably identical) mentions that it's very hard to see on high resolution screens because of how small it is (https://www.pling.com/p/999623/).

The Breeze cursor theme's size is adjustable from System Settings - the dropdown is right in between the header for "Cursors" and the Configure Launch Feedback button. Are you able to access that and change the size from there?

Just for reference, below are some links describing ongoing work around cursor sizes, some of which may have impacted what you've seen between 6.0 and 6.2 already - so there definitely have been and are some challenges with getting cursors correctly sized, just want to make sure in this case if it's something within a particular theme or something controllable more systematically.

https://blogs.kde.org/2024/10/09/cursor-size-problems-in-wayland-explained/
https://bugs.kde.org/show_bug.cgi?id=459161

Thanks,

How does this affect DPI scaling? I understand that the cursor size is tiny when wanting to do additional user directed scale, but without that, shouldn't the cursor be scaled up by 150% with 150% DPI scale, 175% with 175% DPI scale, etc. no matter what the theme does (by simply blowing up the bitmap)? I think this would be the expected behavior.

In case I worded it too vaguely: I think as both a user and a theme creator the base expectation is e.g. a 32x32 cursor bitmap should as a starting point always have the same size relative(!) to the other DPI scaled elements, like the task bar. Then whether a theme allows additional sizes seems like a separate concern once that basic behavior is met.

I'm no personal expert, but for me reading through the blog explaining cursor size challenges (and the subsequent blogs linked from there) helped it click for me of why KWin simply taking a cursor bitmap and multiplying it by a scaling factor doesn't work in a clear/simple way in a system using the Wayland protocol (tl;dr it's the apps that are handling the cursor, so the approach has to be able to be implemented across app toolkits)

There's quite a bit of developer discussion ongoing on this, FWIW: see https://invent.kde.org/plasma/kwin/-/issues/247  https://invent.kde.org/plasma/kwin/-/issues/241  https://invent.kde.org/plasma/kwin/-/issues/187

I think perhaps where we're both getting stuck here, so to speak, is that this bug is originally stating that the cursor size from one configuration, with one scaling factor, should match the cursor size in another configuration with a different scaling factor (in the original report, screenshots 1 and 2 are in a pretty different system scaling factor than 3 and 4), and all were with a cursor theme set that doesn't seem to be set up for scaling to HiDPI screens.

Are you observing that, on your Plasma 6.2.4 device, the Breeze cursor theme is not using a larger size when you change the system scaling? If it's not, that feels like a tangible bug that could be actionable and could have clear progress made?

Otherwise, my guess is this probably doesn't have a solution outside of the broader "re-thinking about how cursors work in the modern Wayland/FreeDesktop world" work and what KWin might need to do that was mentioned in https://bugs.kde.org/show_bug.cgi?id=451158

> Are you observing that, on your Plasma 6.2.4 device, the Breeze cursor theme is not using a larger size when you change the system scaling?

The breeze theme cursor certainly looks too small by default compared to Plasma 6.0.x, see https://bugs.kde.org/show_bug.cgi?id=496861#c5 . Whether that's a bug in cursor rendering or just a bad default setting, I'm not sure.

> all were with a cursor theme set that doesn't seem to be set up for scaling to HiDPI screens.

If that's really part of the problem here as well, I would propose the expected effect should be that this makes the cursor blurry, not thatt it makes the cursor smaller or available with reduced available sizes only. If you compare this to web browsers for example, low DPI images will be blurry and not suddenly shrink. If this needs to be solved in multiple toolkits, perhaps in the long term it would make sense if some library handles this, libcursordpi or something. I realize this is all armchair advice however, but from the outside with the many size bugs and the size restrictions it doesn't really seem like it's working as most users would expect it.

My apologies if this input isn't helpful.

Sorry for the comment spam, but https://bugs.kde.org/show_bug.cgi?id=451158#c12 seems to also suggest there should be pixmap scaling as some sort of fallback. That's basically also what I would have expected as a user.

(In reply to Ellie from comment #11)
> If that's really part of the problem here as well, I would propose the
> expected effect should be that this makes the cursor blurry, not thatt it
> makes the cursor smaller or available with reduced available sizes only.

I might be wrong, but I think that's the crucial point of this one - unless you'd object, I edited the title of this report to reflect that it's more about asking KWin to do things that currently aren't feasible without creating other issues, like blurry cursors. Makes sense that that would be something asked for, I just think it hasn't come up as much because the majority of folks are using the built-in cursor theme that now handles the situation generally correctly.

Created attachment 176471 [details]
Screenshot 5 (Plasma 6.2.5 but suddenly different cursor size at same DPI with same 32px cursor t heme, compare Screenshot 3!)

I appreciate you helping me to file this constructively.

However, after some more testing now on 6.2.4, I suddenly get the "correct"(?) cursor size from 6.0.5, but at the same DPI scaling with the same cursor theme with the same 32px only cursor, and it's clearly different than what I saw before. Check Screenshot 5 and Screenshot 3 for yourself, both use "Classic Flat White" cursors and you can see from the scaling of Firefo and the task bar that it seems to be the same DPI.

Also, when I scale up the DPI, the cursor already gets scaled up and becomes blurry.

Therefore, I'm concluding 1. the feature seems to be already there, but 2. it doesn't seem to be working as intended. My apologies if I'm just confusing something.

Created attachment 176472 [details]
Screenshot 5 (Plasma 6.2.5 but suddenly different cursor size at same DPI with same 32px cursor t heme, compare Screenshot 3!)

Since this seems to be a legit bug and not a wishlist item or feature request, would it be possible to revert it? I don't seem to be able to. Sorry for taking up your time.

I think someone closer to KWin development would need to make calls on this one anyway (including confirming that it's intended to be actively worked on), since I'm interpreting it as a desire for KWin to be able to adapt for things that conceptually should be, but aren't, present elsewhere in the software stack (ex. cursor themes that are missing sizes, toolkits that don't handle HiDPI consistently), which then seems like a feature request. But my understanding is very surface-level, so don't take my triage work as something set in stone.

Sorry for perhaps coming across as pushy but please re-read this comment: https://bugs.kde.org/show_bug.cgi?id=496861#c14

There are different cursor sizes appearing on the plasma panel for the same screen resolution, same DPI scale, same cursor theme, same cursor theme sizes. Also, sometimes on Firefox it is tiny and sometimes it isn't. This can't be intended behavior, can it?

Some observations and a data point

We need to focus on Plasma 6.2.4 as code has changed from earlier versions
We also need to focus on Breeze (vs 3rd party themes), since we have no control over 3rd party cursors (if there is a desire to support scaling for 3rd party cursor themes that don't have all appropriate sizes, that is indeed a feature request)

I tested with Plasma 6.2.4 (Breeze cursor theme) as well as git-master (Oxygen Blue), both on Solus, both cursor size 24
Using text on the screen as a rough visual measure, I changed the display to 150% scale, and then let it revert to 100%
I observed that the cursor was scaled up and down according to the display scale, it seemed correctly proportional to me
The cursor doesn't change size if it's over the panel or Firefox

I also tested with Breeze Light (white cursors) with a built in size of 36. Again, I saw the cursor scale proportionally with the display scaling.

As an aside, testing the Classic Flat White cursor theme, there is no ability to select the size in System Settings - Cursors. It does indeed look smaller than other cursors at 24px. That's a limitation of that cursor set, which again, is 3rd party. For that theme, you can try reaching out to the author. I see a homepage for it in Download New Cursors.

To summarize, n Plasma 6.2.4 with the Breeze cursor theme, you are able to set the cursor size you find acceptable. 
If you want Plasma to do something with cursor themes that don't support HiDPI or are limited to one size, that is indeed a feature request.
The KWin devs can chime in with further detail.

> I saw the cursor scale proportionally with the display scaling.

Sorry if I'm missing something but this seems to be incorrect as being always true, see Screenshot 3 & Screenshot 5. Both are from Plasma 6.2.4 at different DPI sizes, yet the cursor size differs. This is what I discovered in https://bugs.kde.org/show_bug.cgi?id=496861#c14 .

Also, I'm now seeing the tiny cursor on Firefox from Screenshot 6.0.5 but also on 6.2.4. It worked before, but it seems under some conditions this isn't the case.

Sorry, I meant to say "same DPI sizes". You can see with the Plasma taskbar scale, it's the same on Screenshot 3 & 5.

Perhaps it is best if I open a separate issue since this one seems to have gotten a bit off track. But it seems to be a bug, and it's still present in 6.2.4. In any case, sorry for the derailing and confusion.

While I don't object to this feature request and I think it's a good idea, before talking about what cursor sizes are useful it  perhaps would make sense to look at when they unintentionally differ first. Since this ticket derailed a bit, I refiled here trying to limit it to Plasma 6.2.4 and clearly list the situations that are suspicious: https://bugs.kde.org/show_bug.cgi?id=498387 I hope that is productive. Perhaps it makes sense to close this one here, and then refile it once cursor size are stable if there's still a need? Sorry if that's not a useful suggestion.

>perhaps would make sense to look at when they unintentionally differ first. S

It is unintentional. 

There is nothing we can do at a kwin level when the client says "this is my cursor as a pixmap (picture), use it". We don't know whether it happens to be breeze or something super custom, like a crosshair in an RPG video game.

We are making headway in fixing other toolkits, there's no easy fix at a kwin level that we can safely do, sorry

It does seem to affect the plasma task bar too under some circumstances. So there seems to be some potential plasma bug involved, or at least as a naive user from the outside it seems a little like that to me.