Natural, smooth and zoom-relative scrolling

[allejok96]

Scrolling overhaul - a.k.a please make zooming usable for me 😿

• Handle smooth scrolling trackpads and mice gracefully (supersedes #6220)
• Only allow macOS natural scrolling in places where it makes sense
• Scrolling in editors is relative to zoom level
• Minimum scroll speed of Knob and Fader is at least 1% per wheel step
• Only scroll MDI area on empty space or scrollbars
• Discard horizontal scroll in places where it shouldn't be
• Fix bug when scrolling on empty MidiClip
• Remove shift+scroll in favor of Qt's built-in alt+scroll 😱 (see #5169)

Much of the above is done through the use of a single helper function, inspired by Qt's docs and internal code.

knob->incValue(verticalScroll(event, 10));  // scrolling increases knob by 10

Fixes maybe?

• Fixes Window Scrollbar Scrolls Performed With Mouse Wheel Scrolls Entire LMMS Window Scrollbar #3154
• Trackpad sensitivity for knobs and spinboxes #5511
• LCD Widgets reduce value by 2 when finishing trackpad "wheel scroll" action #3370
• Provide a gap between the Note Velocity/Pan toggling and the Volume/Pan bars in Volume/Pan editor of Piano Roll #6413
• Natural scrolling is backwards for knobs, spinboxes #5507
  • Doesn't work for me under Gnome/X11, hope it works for macOS

Things to test

• Scrolling in editors + modifier keys
• Scroll speed of editors at different zoom levels
• Scrolling on widgets like combo box
• Scroll speed of knob and fader
• Natural scrolling on Mac
• Touch pad scrolling

[Veratil]

Where are we calling it multiple times? I don't seem to see any.

[PhysSong]

For diagonal scroll events, Qt 5 generates two events(one for vertical, the other for horizontal) for Qt 4 compatibility. AFAIK, this isn't the case in Qt 6.

[allejok96]

Thanks for the insight @PhysSong, I was wondering about that. This is not a problem atm, but I was thinking about fool proofing. I tried to make it clear in the doc string, but you know.... people. If called improperly the counter will be double increased and it will be hard to detect.

[Veratil]

I'm not a huge fan of these being static. The value will carry over to another event and might cause unintended scroll values.

[PhysSong]

I'm not sure, but per-widget counters might be useful for fractional parts.

[allejok96]

I've considered this a lot... My conclusion is: does it really matter?

Per-widget counter - the logical solution:

1. Scroll knob A 90% of a step - nothing happens.
2. Scroll knob B.
3. Scroll knob A slightly - it turns over.
4. User is surprised, because it's impossible to keep track of the scroll remainder of every knob.

Global counter - the simple solution:

1. Scroll knob A 90% of a step - nothing happens.
2. Scroll knob B slightly - it turns over.
3. User is surprised, because it's impossible to keep track of the scroll remainder of every knob.

Another thing: the remainder is reset every time the wheel changes direction. This is how Qt handles it and it's a clever way to prevent offsets building up over time and also pretty logical when you think about it. But it comes with this side effect:

1. Scroll up 90% of a step - nothing happens.
2. Scroll down 10% - nothing happens.
3. Scroll up 90% - nothing happens.

[tresf]

@allejok96 it's very nice to now have the trackpad on MacOS capable of changing knobs!

One initial observation though is clicking and dragging UP/DOWN raises and lowers them, however swiping UP/DOW on the track pad lowers and raises them. Is this intended? Note, MacOS uses "natural" scrolling by default, and I currently have this enabled.

[tresf]

Another issue... prior to this, on Mac, holding Shift would allow the scroll wheel to switch from up/down scrolling to left/right scrolling. This PR seems to have removed that and I can't figure out a replacement key(s).

[allejok96]

@tresf changing to allowNatural = false in these two places should fix the issue of natural scrolling on knobs. I'll get around to fixing it later...

[tresf]

Wouldn't this change belong in Knob.cpp instead? Anyway, I tried it and it has no effect.

Possibly related: #5507 (comment)

[allejok96]

😢 Could you add a printf somewhere to see if QWheelEvent->inverted() is true? On Linux it's not detected. If we can't know if the scroll is natural or not I guess the only way is having a setting for it.

[tresf]

Sure, if you don't mind being more specific; there's no (obvious) place to add...

printf(QWheelEvent->inverted());

[tresf]

I think I got it...

         model()->incValue(verticalScroll(we, scrollFactor) - horizontalScroll(we, scrollFactor));
 
 
+        printf("Inverted: %d\n", we->inverted());
 
         s_textFloat->setText( displayValue() );

Shows:

Inverted: 1
Inverted: 1
Inverted: 1
Inverted: 1
Inverted: 1
Inverted: 1
Inverted: 1
Inverted: 1

[allejok96]

I hope the last commit fixed this

[allejok96]

Now the functions should handle the differing Alt modifier behavior of macOS and Windows. I hope my use of QFlags isn't awful.

[allejok96]

Fellow devs, I have decided to take a break from coding to pursue other things because it's been getting too much lately and apparently my life balancing skills sucks. I'll just leave the PRs open and if anyone feels like it they can pick it up from there. Sorry, I know it's abrupt but sometimes you have to be ❤️

[luzpaz]

Anyone able to pick up the mantle on this ?

[regulus79]

Is this PR still active?

I apologize, I may have accidentally worked over you by implementing scrolling relative to zoom level in #7476. But this PR looks way cleaner and more general, so please feel free to discard my changes.

[tresf]

Is this PR still active?

I apologize, I may have accidentally worked over you by implementing scrolling relative to zoom level in #7476. But this PR looks way cleaner and more general, so please feel free to discard my changes.

The OP took a break so no worries. This PR has over a dozen file conflicts, so a new PR should probably be used over this one.