3.0.0 Beta Feedback

[noah-nuebling]

The Mac Mouse Fix 3 Beta is here!

The best place to share a specific problem or idea is Feedback Assistant.

But you can also use this discussion to quickly share any thoughts or questions.

[CavalloneChen]

Will the code base stay open-sourced after MMF 3? I'm willing to pay to support your efforts, but would also like to see the project stay as open as possible.

[noah-nuebling]

That's really cool @krackers! I'm super stoked to see people put effort thought into the app. I'd also be really interested to try out your upgrades if want to share them!

[krackers]

Oh the version I cobbled didn't use mac-mouse-fix as a base (since I was working on it before I knew MMF existed), so it's its own standalone thing. I'll post the source below, but honestly I don't think it's of much use to anyone since the code is ugly with tons of global state and everything is hardcoded to my preferences. Basically it implements the following features

• Use button 3 + mouse movement for scrolling (and cursor stays locked when scrolling)
• Use button 3 + right click + mouse movement for fluid zoom
• Use command + button 3 + mouse movement for fluid zoom
• Fluid zoom has events weighted so things feel linear (without the weighting, I was getting wonky behavior, but MMF didn't seem to do this so idk)
• Scrolling direction is determined based on cumulative sum of last 10 event vectors (inspired by the drag smoothing in MMF 3)
• Can ignore for certain apps (currently not used though)

#import <ApplicationServices/ApplicationServices.h>
#import <Foundation/Foundation.h>
#include <AppKit/AppKit.h>

#pragma GCC diagnostic ignored "-Wdeprecated-declarations"

#define SCALE 3
#define BUTTON 3

//#define LOGGING

#ifdef LOGGING
#   define NSLog(...) NSLog(__VA_ARGS__)
#else
#   define NSLog(...) (void)0
#endif


int idx = 0;
int dxCum = 0;
int numSamples = 0;
int dxArr[11];
int dyArr[11];
int dyCum = 0;
CGPoint savedPoint;

enum {
    kIOHIDEventPhaseUndefined                               = 0x00,
    kIOHIDEventPhaseBegan                                   = 0x01,
    kIOHIDEventPhaseChanged                                 = 0x02,
    kIOHIDEventPhaseEnded                                   = 0x04,
    kIOHIDEventPhaseCancelled                               = 0x08,
    kIOHIDEventPhaseMayBegin                                = 0x80,
    kIOHIDEventEventPhaseMask                               = 0xFF,
    kIOHIDEventEventOptionPhaseShift                        = 24,
};

typedef uint16_t IOHIDEventPhaseBits;


void postMagnificationEventWithMagnification (double magnification, IOHIDEventPhaseBits phase)
{
    // Using undocumented CGEventFields found through Calftrail TouchExtractor and through analyzing Calftrail TouchSynthesis to create a working magnification event from scratch
    
    CGEventRef event = CGEventCreate(NULL);
    CGEventSetType(event, 29); // 29 -> NSEventTypeGesture
    CGEventSetIntegerValueField(event, 110, 8); // 8 -> kIOHIDEventTypeZoom
    CGEventSetIntegerValueField(event, 132, phase);
    CGEventSetDoubleValueField(event, 113, magnification);
    CGEventPost(kCGHIDEventTap, event);
    CFRelease(event);
}

bool startedzoom = false;
bool scrolling = false;

int cumZoom = 0;

void handleMagnificationScrollWithAmount(int ticks) {
    if (!startedzoom) {
        startedzoom = true;
        postMagnificationEventWithMagnification(0, kIOHIDEventPhaseBegan);
        // Seems to be needed for some reason since otherwise there's a lag
        postMagnificationEventWithMagnification(0, kIOHIDEventPhaseChanged);
    }
    cumZoom += ticks;
    int multp = 1;
    if (abs(cumZoom) < 3 && ticks < 3) {
        multp = 20;
    } else if (abs(cumZoom) < 5 && ticks < 3) {
        multp = 10;
    } else if (abs(cumZoom) < 20 && ticks < 3) {
        multp = 8;
    } else if (abs(cumZoom) < 30 && ticks < 3) {
        multp = 4;
    }
    //NSLog(@"cumZoom %d\n", cumZoom);
    ticks *= multp;
    postMagnificationEventWithMagnification(-1*ticks/300.0, kIOHIDEventPhaseChanged);
}

void endzoom() {
    if (startedzoom) {
        startedzoom = false;
        postMagnificationEventWithMagnification(0, kIOHIDEventPhaseEnded);
        cumZoom = 0;
    }
}

bool rightMouseDown = false;
bool modifierDown = false;
bool otherButtonDown = false;


NSRunningApplication* runningApplication;
int shouldIgnore = false;
bool shouldZoom() {
    return (modifierDown) || (rightMouseDown && otherButtonDown);
}

int simulatedRightClicksPending = 0;
CGEventRef cgEventCallback(CGEventTapProxy proxy, CGEventType type, CGEventRef event, void *refcon)
{
    switch (type) {
        case kCGEventRightMouseDown: {
            if ( CGEventGetIntegerValueField(event, kCGMouseEventSubtype) == 3 /* multitouch */) {
                break;
            }
            rightMouseDown = true;
            bool isSimulatedEvent = CGEventGetIntegerValueField(event, kCGEventSourceUserData) == 0xf00;
            
            if (otherButtonDown) {
                return NULL;
            } else if (!isSimulatedEvent) {
                NSLog(@"Physical Right mouse down\n");
                CGEventRef copy = CGEventCreateCopy(event);
                CGEventSetIntegerValueField(copy, kCGEventSourceUserData, 0xf00);
                simulatedRightClicksPending++;
                dispatch_after(dispatch_time(DISPATCH_TIME_NOW, (int64_t)(50 * NSEC_PER_MSEC)), dispatch_get_current_queue(), ^{
                    if (!otherButtonDown) {
                        CGEventPost(kCGHIDEventTap, copy);
                    } else {
                        NSLog(@"HIT SPECIAL CASE");
                    }
                    CFRelease(copy);
                });
                return NULL;
            } else {
                simulatedRightClicksPending--;
                NSLog(@"Sent Right mouse down event\n");
            }
        }
            break;
        case kCGEventRightMouseUp: {
            if ( CGEventGetIntegerValueField(event, kCGMouseEventSubtype) == 3 /* multitouch */) {
                break;
            }
            
            bool isSimulatedEvent = CGEventGetIntegerValueField(event, kCGEventSourceUserData) == 0xf00;
            
            if (simulatedRightClicksPending > 0 && !isSimulatedEvent) {
                NSLog(@"Mouse up during simulated period... delaying\n");
                CGEventRef copy = CGEventCreateCopy(event);
                CGEventSetIntegerValueField(copy, kCGEventSourceUserData, 0xf00);
                dispatch_after(dispatch_time(DISPATCH_TIME_NOW, (int64_t)(50 * NSEC_PER_MSEC)), dispatch_get_current_queue(), ^{
                    CGEventPost(kCGHIDEventTap, copy);
                    CFRelease(copy);
                });
                return NULL;
            } else {
                NSLog(@"Sent mouse up event\n");
                rightMouseDown = false;
            }
            
        }
            break;
        case kCGEventOtherMouseDown:
        {
            NSLog(@"Other mouse down\n");
            int x =(int32_t)CGEventGetIntegerValueField(event, kCGMouseEventButtonNumber);
            if (x == 3 || x == 4) {
                otherButtonDown = true;
                idx = 0;
                numSamples = 0;
                dxCum = 0;
                dyCum = 0;
                savedPoint = CGEventGetLocation(event);
                
                return NULL;
                // Suppress event
                
            }
        }
            break;
            
        case kCGEventOtherMouseUp:
        {
            NSLog(@"Other mouse up\n");
            int x =(int32_t)CGEventGetIntegerValueField(event, kCGMouseEventButtonNumber);
            if (x == 3 || x == 4) {
                otherButtonDown = false;
                idx = 0;
                numSamples = 0;
                dxCum = 0;
                dyCum = 0;
                scrolling = false;
                endzoom();
                // Suppress event
                return NULL;
            }
        }
            break;
            
            
        case kCGEventOtherMouseDragged:
        case kCGEventRightMouseDragged:
            //NSLog(@"Mouse dragged\n");
            if (!otherButtonDown) {
                break;
            }
            
               // printf("Enabled");
            int32_t deltaX = (int32_t)CGEventGetIntegerValueField(event, kCGMouseEventDeltaX);
            int32_t deltaY = (int32_t)CGEventGetIntegerValueField(event, kCGMouseEventDeltaY);
            
     
            if (!shouldZoom()) {
                if (numSamples < 10) {
                    dxCum += deltaX;
                    dyCum += deltaY;
                    dxArr[idx] = deltaX;
                    dyArr[idx] = deltaY;
                    idx = (idx + 1) % 10;
                    numSamples++;
                } else {
                    dxCum -= dxArr[idx];
                    dyCum -= dyArr[idx];
                    dxCum += deltaX;
                    dyCum += deltaY;
                    dxArr[idx] = deltaX;
                    dyArr[idx] = deltaY;
                    idx = (idx + 1) % 10;
                    
                    CGEventSourceRef source = CGEventSourceCreate(kCGEventSourceStateCombinedSessionState);
                    if (5 * abs(dyCum) > abs(dxCum)) {
                        deltaX = 0;
                    } else {
                        deltaY = 0;
                    }
                    //NSLog(@"%d %d\n", deltaX, deltaY);
                    CGEventRef scrollEvent = CGEventCreateScrollWheelEvent(source, kCGScrollEventUnitPixel, 2,
                                                                           -SCALE * deltaY, -SCALE * deltaX);
                    scrolling = true;
             
                    
                    CGEventPost(kCGSessionEventTap, scrollEvent);
                    
                    CFRelease(scrollEvent);
                    CFRelease(source);
                }
            } else {
                handleMagnificationScrollWithAmount(deltaY);
            }
            
            CGWarpMouseCursorPosition(savedPoint);
            CGAssociateMouseAndMouseCursorPosition(true);
          
            break;
        
        case kCGEventFlagsChanged: {
            CGEventFlags flags = CGEventGetFlags(event);
            modifierDown = flags & kCGEventFlagMaskCommand;
        }
            break;
            
        default:
            break;
    }
    
    return event;
}


int main(void)
{
    CFMachPortRef eventTap;
    CFRunLoopSourceRef runLoopSource;
    
 CGEventMask mask = // kCGEventMaskForAllEvents;
    CGEventMaskBit(kCGEventOtherMouseDown) | CGEventMaskBit(kCGEventFlagsChanged) | CGEventMaskBit(kCGEventOtherMouseUp) | CGEventMaskBit(kCGEventOtherMouseDragged) | CGEventMaskBit(kCGEventRightMouseDown) | CGEventMaskBit(kCGEventRightMouseUp) | CGEventMaskBit(kCGEventRightMouseDragged);
    
    eventTap = CGEventTapCreate(kCGSessionEventTap, kCGHeadInsertEventTap, 0, mask, cgEventCallback, NULL);
    if (eventTap == nil) {
        NSLog(@"FAILED TO CREATE EVENT TAP");
        return -1;
    }
    runLoopSource = CFMachPortCreateRunLoopSource(kCFAllocatorDefault, eventTap, 0);
    
    CFRunLoopAddSource(CFRunLoopGetCurrent(), runLoopSource, kCFRunLoopCommonModes);
    CGEventTapEnable(eventTap, true);

    [NSWorkspace.sharedWorkspace.notificationCenter
        addObserverForName:NSWorkspaceDidActivateApplicationNotification
        object:nil
        queue:nil
        usingBlock:^(NSNotification* notification) {
                runningApplication = notification.userInfo[NSWorkspaceApplicationKey];
                if ([runningApplication.bundleIdentifier isEqualToString:@"com.apple.ScreenSharing"]) {
                    shouldIgnore = true;
                } else {
                    shouldIgnore = false;
                }
                NSLog(@"%@", runningApplication);
        }];
    CFRunLoopRun();
    
    CFRunLoopRun();
    
    CFRelease(eventTap);
    CFRelease(runLoopSource);
    
    return 0;
}

[noah-nuebling]

I'll try it out sometime. Thanks a lot for sharing @krackers!!

[noah-nuebling]

@krackers, I just tested it out, it's really cool, thanks again for sharing!

It feels really natural to use the dragging to zooming. And it's interesting how you're ignoring clicks from a trackpad, that's a good idea. I also like switching between the dragging and the zooming by just clicking or releasing the Right Mouse Button.
Why are you delaying the simulated right clicks by 50 ms?

[krackers]

@noah-nuebling Thanks for trying it out!

I also like switching between the dragging and the zooming by just clicking or releasing the Right Mouse Button.

Yup this is exactly why I added that feature, so I can easily switch between scrolling and zooming.

Why are you delaying the simulated right clicks by 50 ms

If I didn't delay the right click, then in order to zoom (from a state when I wasn't already scrolling) I'd have to first hit button 3 then hit right click (otherwise if I ended up hitting right click first, it would trigger the context menu). Ideally I'd like to just mash button 3 and right click "simultaneously" though, so adding the 50ms delay means that the exact order of button triggering doesn't matter. I tried experimenting with something where if I hit button 3 after right click it would dismiss the context menu, but that didn't work. If there's a better way to achieve the same thing do let me know though!

---

I'm also not 100% satisfied with the smoothing algorithm. It works fine for large movements, but for small movements (e.g. scrolling just 1 line) it sometimes interprets as vertical when I meant horizontal or vice-versa. This isn't a fault of the smoothing per se but probably because I'm really sloppy when scrolling and often move the trackball at a 45deg angle. I've been trying it with the dXcum and dyCum updates in the numSamples >= 10 block commented out, to effectively freeze the scrolling direction after the initial 10 samples, which makes things more predictable at the expense of needing to release whenever changing direction. I kind of like this better because I can be as sloppy as I like without affecting things.

[rmateu]

Hi @noah-nuebling ! very excited about all the new features and the prospect of MMF getting more continued updates.

My only feedback is: increase the price:

• MMF is critical enough for me that I'm price inelastic to $9.99 easily.
• If that's crazy, then $4.99 makes little difference (yet it's 150% more for you)
• If you really wanna keep "cheap coffee" pricing, and not go the "subscription" route, maybe a pricing model based year of updates (like Hook, Sketch, etc)
  • $1.99 for all the updates in 12 months. Afterward you can upgrade again or simply stay with the funcionality you bought + 12 months.

Your probabably have stats of # downloads, etc. But my thinking is: I already bought a 3rd party mouse that costs $75+, MMF gives me functionality that makes the purchase worth it. Also, it's easier to lower price than increase it.

In any case, take my money!

[noah-nuebling]

In any case, take my money!

That's great to hear, hahaha! I'm really glad you appreciate Mac Mouse Fix so much and that you put so much thought into what's best for the project and for me! Very sweet.

My thinking behind the pricing was that I want as many people as possible to use the app and that I want those users to be really happy with the app and with their purchase.
I'd rather see the project grow than my wallet.
And I think I should still be able to make a living at $1.99 so I can work on the project full time, which is also important.

MMF is critical enough for me that I'm price inelastic to $9.99 easily.

I think $9.99 would be the absolute maximum. I've paid more for similar software and it always left a bit of a bad taste in my mouth.

If that's crazy, then $4.99 makes little difference (yet it's 150% more for you)

You're totally right. Even selling at $2.99 would almost double the revenue, and it's such a small difference to the consumer. So I'm definitely keeping it an option to increase the price. I think I'll release at $1.99 and maybe increase the price slowly if it doesn't make enough revenue to find a sweet spot and see how people react. I feel like people would be totally fine with that. Or do you have different thoughts or experiences here since you said "it's easier to lower price that increase it"?

If you really wanna keep "cheap coffee" pricing, and not go the "subscription" route, maybe a pricing model based year of updates (like Hook, Sketch, etc)

I personally really dislike subscriptions as a consumer. I'll only purchase a subscription if I really have to. So that's something I definitely want to avoid.
But as much as I dislike subscriptions from a consumer perspective, I can see why it's really attractive to so many companies and creators to have a continuous revenue stream from their existing userbase. They don't have to grow and grow just to stay alive which is great.

I think your idea about a pricing model where you buy a year of updates is an interesting middle ground where the creator still has a stream of revenue from the existing userbase, but the userbase also has something they own after spending money, which is really nice.

How do you feel about a pricing model where you get all updates until the next major feature release? So you'd get MMF 3 forever plus all updates until MMF 4?

Again, thanks so much for your kind words and ideas!

And of course I'm really interested to hear everyones' thoughts and feelings on this! As I said I think monetization is a really important step for the project and I want to make it as good as possible and you guys to be happy with it.

[nulaft]

I would like to try the v3 beta, but for some reason I cannot get it to run. I have tried right-click open and that xattr command without success.

[noah-nuebling]

Hey @nulaft, thanks for posting! Does this version work for you?

[nulaft]

Yes, that version works for me. Like others have said, it also crashes for me when clicking on the enable toggle, but clicking on the edge of it worked for me as well so I can give it a try :)

[noah-nuebling]

Great!

[andreadaoud]

A use case on per-app disable of Mac Mouse Fix: When using Microsoft Remote Desktop to connect to remote PC, we want the scrolling event to be passed directly to the PC we are connecting. Enabling Mac mouse fix in this scenario will make scroll unusable on remote Windows.

[noah-nuebling]

Hey @andreadaoud! In case you can use MMF 3 at the moment, does going to the Menu Bar Item and disabling MMF for Scrolling make it usable with Remote Desktop? If so do you think that's an adequate solution?

[andreadaoud]

This works, but require manual operation, which I think is less ideal. Anyway, thank you for your great work!

[noah-nuebling]

Ok let me know if it bothers you too much or if you have more thoughts and feedback on that. And thanks for your kind words, I'm glad you like the app!

[attaboyBrad]

I had a strange edge case with Mac Mouse Fix 2: I switch between a mix of pointing devices simultaneously and the Touch-scroll speeds on my Wacom tablet were unusably fast if I was to keep the mouse-scroll-wheel speeds reasonable. I used app-specific settings to turn off one of the settings (whichever one was leftmost) in the apps I used the Wacom with primarily, but it was still a daily annoyance.

Not sure if you fixed my issue on purpose, but it appears to be fixed with v3.

I am so grateful for MM2 that MM3 was an insta-buy regardless of this fix, but I'm so grateful. Enjoy the milkshake. Shoot--I'm gonna go buy you another one... Done.

Possible feature request: a way to select the scroll direction individually per device. It seems like something like Karabiner should make this easy, but I can't seem to get it sorted. I actually sought out beta versions of MM2 as part of this quest--thus my presence in this thread.

At risk of sounding like a lunatic: I have a ShuttleXpress, Magic Trackpad, Razer Mouse, Wacom Pro tablet with pen+touch support, and an Audient USB audio interface that lets me use the volume knob as a scroll wheel--and I switch between them all over the course of the day. I'm having a devil of a time getting them to match direction. I think its because the ShuttleXpress uses a custom driver or something so may not be a "mouse" issue at all as far as MacOS is concerned.

Another suggestion: in addition to the "Milkshake Donation" link in the "about" page you might also include a link that lets people Gift licenses to friends.

Anyway, thanks for MM3.

[noah-nuebling]

Thanks so much for your thoughts and for the milkshakes @attaboyBrad :)

Not sure if you fixed my issue on purpose, but it appears to be fixed with v3.

I made some changes to try and ignore drawing tablets in the 2.2.1 Beta that are also present in the 3.0.0 Beta, glad to hear it actually works!

Possible feature request: a way to select the scroll direction individually per device

It's a long way off at this point, but I do plan to add better support for devices with special hardware features like horizontal scrollwheels etc. and I do plan to add some amount of device-specific settings at this point. This is a pretty compelling reason to add per-device scroll inversion. I didn't think about this so thanks for making me aware!

At risk of sounding like a lunatic

Everyone's a little crazy :)

... include a link that lets people Gift licenses to friends

Ooh that's really cool! Also hadn't been thinking about that, but I like it! I'm not sure how easy this is to implement, but I'll give it some thought, this would definitely be a cool addition!

[carlfriedrich]

Hey @noah-nuebling, thanks for releasing Beta 3 and putting so much valuable effort into it!

I have another use case for app-specific settings: when using audio DAW software Cubase, cmd + scroll wheel for zooming does not work when it's enabled in MMF. I have to disable "cmd-scroll to zoom" to make it work in that app. Seems totally counter-intuitive, and I really don't know why this is the case, but that's the only way I get it working.
Since I have Cubase open on my Mac all day and constantly am switching between Cubase and other apps, deactivating it manually via the MMF menu bar icon is not an option for me, so I am gonna stick with v2 until you bring back app-specific options.

Or do you have an idea why Cubase won't zoom correctly with MMF zooming enabled? Maybe it's because Cubase does not support pinching and has implemented its own cmd+zoom feature which gets overridden by MMF? I that's the case, I assume there is no other way to make it work, is there?

Another thing I noticed in v3: when I disable the modifier key for zooming completely, Cubase zooming works inverted (cmd+wheel up zooms OUT instead of IN). So in v3 the only way to make zooming in Cubase work really is to disable zooming in MMF completely via the menu bar icon.

[noah-nuebling]

Hey @carlfriedrich, thanks a lot for your feedback!

I think you're right about your speculation on why Cubase zooming doesn't work with Mac Mouse Fix. Cubase just sees "someone is pinching on a Trackpad" when you Command-Scroll with Mac Mouse Fix enabled. And if it doesn't have pinch-to-zoom support, then that won't do anything. The best solution to this would be Cubase adding support for pinch-to-zoom on a Trackpad.

Until then, the best solution is probably to turn off Mac Mouse Fix scrolling when you're using Cubase. If it's too annoying for you to turn off Mac Mouse Fix from the menubar every time, you can disable Command-Scroll to Zoom entirely as you already discovered.
The issue about the zoom direction being inverted is weird. I would guess it's because MMF 3 simulates Trackpad scrolling more accurately than MMF 2. So Cubase thinks you're scrolling on a Trackpad and inverts the zooming direction for some reason. You should be able to fix this by turning Smooth Scrolling off entirely. Then, Cubase should think you're scrolling on a normal mouse, not a Trackpad and it hopefully won't invert the scrolling direction anymore.

After that, I don't really have any good ideas how to make your experience better. Aside from bringing back App-Specific settigns. Which i hope I'll be able to soon!

Hope that's a little helpful! Let me know if you have moe questions.

[carlfriedrich]

The best solution to this would be Cubase adding support for pinch-to-zoom on a Trackpad.

Totally agree. Even without using MMF, this would be absolutely useful. It has been an open feature request for years, though. No idea why Steinberg doesn't fix this. It's such a common UI feature (which has been around for roughly 15 years!), how can any major software suite just not support this in 2022?!

However, thanks for your reply! I will try MMF3 again as soon as it has app-specific settings.

[melyux]

Some feedback: I'm seeing a lot higher CPU usage in v3 than v2. Scrolling around on v2 would produce a maximum of 1.0% CPU usage. Doing the same on v3 goes to 3.0% on the same computer. Is there a way to lower the usage for this?

Feedback 2: Scrolling on the v3 beta 3 is still pretty choppy and frequently skips. It's a bit better than v2 though

Question: How much is the impact of Magic Mouse Fix on battery life when a mouse is not being used, only the built-in trackpad? I don't know if I should be disabling Magic Mouse Fix when I'm on the go, since it's constantly running in the background even when not using a mouse.

[noah-nuebling]

Hey @melyux, thanks for your feedback!

I already looked into optimizing the new scrolling algorithms but it's just going to be a bit more CPU intensive now. That's because:

• It's accurately simulating Trackpad scrolls now which is relatively complex, so the scrolling code just needs to do more stuff.
• I used a lot of Swift code for the new scrolling algorithms which is generally a bit slower than the old C code, and when it needs to interface with C code that adds even more slowdown.

But some of the increased CPU usage you're observing is also explained by other things I think:

• The Beta versions are so called "Debug" builds which makes it so the app outputs more info about what's happening when it crashes or when something else goes wrong. This is so that I can see what's going on when you guys file Bug Reports. But "Debug" build also make Swift code incredibly slow. Like 40x slower or something. The final release will be a "Release" build instead of a "Debug" build, so it should be a lot faster by virtue of that.
• Under the default settings, in MMF 3, the scroll animations are a lot longer which causes them to show up as more CPU usage in Activity Monitor even if they are not less efficient.

I just did some testing. Scrolling continuously under MMF 2 I got a max sustained CPU usage of 2.2%, while under MMF 3, using a "Release" build, I got a max usage of 3.0%. So it should be around 30% slower.

When it comes to the choppiness you observe, I think that might also be due to the longer animations. I experienced the same effect on my 2015 MBP when I had a lot of windows open and the scrolling started dropping frames, MMF 2 seemed a lot less choppy than MMF 3, but that was because MMF 2 has very short animations so you don't notice the dropped frames. When I continously scrolled on MMF 2 it had the same amount of chopiness. And scrolling on the Trackpad also had the same choppiness. Basically, the bottleneck that makes things choppy was never MMF on the machines I tested. It just makes the choppiness more or less obvious depending on how long the animations are.

Does that match your observations?

You can already make the animations shorter in MMF 3 by turning off the "Inertial Scrolling" option. The animations will still be a bit longer than MMF 2, but it should feel a lot less choppy already. Before MMF 3 releases, I want to make the animations even shorter so that MMF 3 feels just as responsive as MMF 2 when you turn the "Inertial Scrolling" option off.

To your question about battery life: The CPU usage when you don't use your mouse should be 0%. It should not be doing anything, and it shouldn't affect your battery at all. If it is doing something, that's a bug. I think I saw a bug in the MMF 3 Beta where it always uses CPU when you move the pointer, maybe you saw that? I'm confident I'll be able to resolve that until release.

Let me know what you think! And don't hesitate to ask if you have more questions!

Edit:

Feedback 2: Scrolling on the v3 beta 3 is still pretty choppy and frequently skips. It's a bit better than v2 though

Did you really mean to say that v3 is less choppy than v2? Or did you mean to say that v3 Beta 3 is less choppy than v3 Beta 2?

Just making sure since I would expect v3 to be more choppy than v2 if anything, since it uses more CPU.

[attaboyBrad]

I'm having an issue with pointing, and that issue started recently--I think roughly around the time I installed the MM3 beta. I don't actually have any better reason to think it stems from MM3, but given that you've developed MM3, perhaps you can point me in the right direction:

What tools do you use to track keyboard/mouse events and movements to see if MM3 is giving the expected outputs given a set of inputs?

Something in my setup seems to be either clicking or tabbing out of text fields (at least, that's when I notice it), and I'm not sure how to go about tracking down the culprit. It's not frequent--just a few times a day--but regular enough to be annoying.

I have a stupid number of computer peripherals and a heap of little utility apps that I run in the background, so I'm hoping there's a way to track down this bug besides weeks of a/b testing every single utility app and peripheral.

I've just now started running Karabiner-EventViewer. I'm hoping you can perhaps recommend a better tool.

To make this comment more directly MM3 related: turning off inertial scrolling (based on your feedback to an earlier comment by melyux) has made a huge difference in my perceived scroll-wheel scrolling responsiveness. It might be worth including some info from your response there as tool tips or something on the various options in MM3 so interested users have a better idea of what they actually do and their possible side effects.

[noah-nuebling]

Hey @attaboyBrad, thanks for reaching out! I'm not sure I can help you with that, but here are some thoughts:

I don't have a better idea for debugging this than the ones you already talked about. Namely, trying to turn off different software and maybe switching between mice to see if anything changes. Looking at an event logger might help but is probably pretty overwhelming. The Karabiner-EventViewer is probably a lot better than this, but I hacked together a basic event logger for fun:

EventLoggerForBrad.zip

It should log every event that is happening on your computer, no matter if those events stem from your physical input devices or apps like Mac Mouse Fix.
Unfortunately it can't tell you which events stem from where. The Apple APIs make this really hard. So I'm not sure if this is too useful for debugging your problem, but you might give it a try.

It's a command-line-tool and you can use it by executing it from the terminal. It should then start writing all the events that are happening on your Mac to the Terminal output.

Here's what the output looks like:

Another thing I can think of is that I had a somewhat similar problem in the earlier Ventura Betas. When I tried to drag and drop something it would drop it before I release the mouse button. This went away in a later Beta though. Maybe Beta 6 or something. Not sure. So if you're also on a macOS Beta maybe it has to do with that?

To your comment on Inertial Scrolling:

I'm glad it's working better for you now! I agree with you it's probably a good idea to add some explanation for the inertial scrolling. I think I might also turn it off by default, since I think most people will prefer the responsiveness over the smoothness and ease of scrolling long distances.

Hope that helps a little! Let me know if you have questions or anything.

---

Oh and here's the source code for the logger if you want to customize it:

//
//  main.m
//  EventLoggerForBrad
//
//  Created by Noah Nübling on 2022/9/22.
//

#import <Foundation/Foundation.h>
#import <Cocoa/Cocoa.h>

CGEventRef eventTapCallback(CGEventTapProxy proxy, CGEventType type, CGEventRef event, void *userInfo) {

    if (type == kCGEventTapDisabledByTimeout || type == kCGEventTapDisabledByUserInput) {
        
        printf("Event tap was disabled due to %s. Quitting.\n", type == kCGEventTapDisabledByTimeout ? "timeout" : "user input");
        exit(1);
    }
    
    NSEvent *e = nil;
    
    BOOL convertibleType = 0 < type && type <= 100000;
    BOOL conversionSuccessful = NO;
    
    if (convertibleType) {
        e = [NSEvent eventWithCGEvent:event];
        conversionSuccessful = e != nil;
    }
    
    NSDate *now = [NSDate date];
    NSDateFormatter *dateFormatter = [[NSDateFormatter alloc] init];
    [dateFormatter setDateFormat:@"HH:mm:ss"];
    NSString *nowStringNS = [dateFormatter stringFromDate:now];
    const char *nowString = [nowStringNS cStringUsingEncoding:NSUTF8StringEncoding];
    
    if (convertibleType && conversionSuccessful) {
        printf("%s - %s\n", nowString, [e.description cStringUsingEncoding:NSUTF8StringEncoding]);
    } else {
        printf("%s - Event of type %d couldn't be displayed\n", nowString, type);
    }
    
    return event;
}

int main(int argc, const char * argv[]) {
    @autoreleasepool {
        
        /// Get runLoop for current thread
        CFRunLoopRef loop = CFRunLoopGetCurrent();
        
        /// Create eventTap
        CFMachPortRef eventTap = CGEventTapCreate(kCGAnnotatedSessionEventTap, kCGTailAppendEventTap, kCGEventTapOptionDefault, kCGEventMaskForAllEvents, eventTapCallback, NULL);
        if (eventTap == nil) {
            printf("Event tap couldn't be created. Make sure this command-line-tool and your Terminal have accessibility permissions.");
            exit(1);
        }
        CFRunLoopSourceRef source = CFMachPortCreateRunLoopSource(kCFAllocatorDefault, eventTap, 0);
        CFRunLoopAddSource(loop, source, kCFRunLoopCommonModes);
        
        /// Run the runLoop
        CFRunLoopRun();
        
        /// Release stuff (probably not necessary)
        CFRelease(eventTap);
        CFRelease(source);
        }
    return 0;
}

[attaboyBrad]

Wow! This is about 10x the response I hoped to get. Thanks so much.

I'm going to see if I can make a button on my StreamDeck produce some unique output in your Logger and try to push that button as quickly as I can after I observe that issue I'm having. Should help me isolate only those events that happened near that time.

[marianmelinte93]

I have some feedback and one possible bug.

Feedback 1: Add buttons to enable/disable Buttons and Scrolling on their respective page as well, and not just in the Menu Bar. Personally, I activated the Menu Bar just to see what I might find there, but I wasn't expecting it to have functionality that cannot be done from the main UI as well.

Feedback 2: Switching to the Buttons or Scrolling pages enables them. This might be intended behavior, linked to feedback above. Considering that the Menu Bar might not be enabled, you'd probably want them to be enabled automatically. This is a bit strange to me, because if I have the Menu Bar active, I wouldn't want to see those options that I changed manually to be changed back automatically.

Feedback 3: In the Buttons page, the section where the buttons can be configured is always scrollable. I was testing some things out and found that even if I removed a lost of options, the entire window would shrink and the scroll would still be there.
Maybe this is intentional, but it looks a bit strange considering that I have enough space to not have a scroll here:

Bug: I don't seem to be able to deactivate the Scrolling part. I use version 3.0.0 Beta 3 (18468) on a MacBook Pro (M1 Pro).
A simple test for me was to leave the Natural Scroll Direction unchecked, which makes the scroll direction opposite from the Mac's default behavior, then I disabled Scrolling, but nothing changed, the scroll direction is the same. If i turn off Enable Mac Mouse Fix entirely, then it stops behaving like this. As far as I could see, not just the Natural Scroll Direction stays active, but this was the easiest for me to explain.
Edit: I also found that disabling and enabling Show in Menu Bar would enable both Buttons and Scrolling, regardless of their previous state. Which again, if we disregard Feedback 1, makes sense if there's no way to enable them without the Menu Bar.

Question: If the bug would be fixed (assuming that there is one, and it's not just me), does this mean that the Scrolling won't be affected at all, or MMF will still be hooked to the system? I'm asking because I used MMF v2 along with LinearMouse and I was able somehow to change the scrolling behavior (using LinearMouse) to one very similar to what I'm used to in Windows (Enable linear scrolling, by lines: 3), and it worked great in remote desktop to Windows machines as well.
Adding this kind of behavior to MMF (linear scrolling, no acceleration) would be amazing, but I would be more than happy to continue using both, as long as they can work together.

Thank for all the work you're doing!

[attaboyBrad]

Thanks so much for mentioning LinearMouse. I took a gander and my jaw dropped when I saw they had per-device scroll settings. Hopefully MM3 builds those in at some point, but in the meantime adding LinearMouse for that feature alone will be a godsend.

[noah-nuebling]

@marianmelinte93, thanks for your detailed feedback and for your encouraging words!

On the menubar item

The toggles in the menubar are meant to quickly turn features off and on if you're trying to play a game or use an app that's incompatible with Mac Mouse Fix. All settings that are available through the menubar are also available through the normal UI. For example, if you just delete all the actions on the Button's tab, that's the same thing as deactivating Mac Mouse Fix for Buttons from the Menu Bar. It's just a lot quicker and easier to do this from the menubar, but it's all possible in the normal UI.

I can see how this can be a little confusing and I might add some hints or comments in the future to make this more clear.

Do you think the UI makes more sense looking at it from that perspective?

On scrollbars

The scrollbars showing up in the Action Table is a bug. I think I've fixed it! Thanks for making me aware.

On LinearMouse compatibility and and linear scrolling

The MenuBar item doesn't affect the scrolling direction. This is on purpose, because it's meant as a compatibility mode with games and other apps, and the scrolling direction doesn't normally cause any compatibility issues as far as I know.

This also means that Mac Mouse Fix stills intercepts the scrolling input from your mouse even when it's deactivated for scrolling from the menubar, and this might cause issues trying to use it together with LinearMouse.

It would be totally feasible and a good change to turn off all scrolling interception when all the scrolling enhancements are turned off and the scrolling direction matches the system scrolling direction. Currently, Mac Mouse Fix will still intercept scrolling input in that scenario, but it won't do anything, except waste a little CPU. So that would definitely be a good change and it should help with Linear Mouse compatibility if there's an issue currently.

As far as building linear scrolling into Mac Mouse Fix, this should definitely be possible, but I'm not sure I could implement it in a great way, since I don't have a good understanding why certain people prefer it. You mentioned that it feels similar to your Windows setup, can you think of other reasons why you like it over other scrolling configurations?

Thanks again for your feedback!

[krackers]

Fwiw to me linear scrolling makes sense when you have a scroll wheel that can be "unlocked" since you already get some amount of acceleration from the physics of the thing. For that reason it also makes sense to me to use it when simulating scrolling using pointer movement (e.g. trackball). Although this is a lot of cases to support at this point ({natural scrolling, inverted scrolling} x {smooth continuous scroll, accelerated scroll, linear scroll} x {real scroll wheel, simulated scroll via cursor}) so just from a perspective of trying to minimize cyclomatic complexity it would seem better to try to separate the concerns and allow it to be handled by different apps.

---

(Tangentially the developer of LinearMouse is also awesome! I remember you once discussed with me how you found a way to get the underlying IOHID event from a CGEvent, turns out the LinearMouse dev also (independently?) discovered the same mechanism and also found some other private IOHID methods you may be interested in: linearmouse/linearmouse#200 (reply in thread))

[marianmelinte93]

@noah-nuebling, thanks for being open for a discussion!

On the menubar item

I've been testing the app a lot that I started doubting myself whether the menubar is active by default or not. I kind of remember it wasn't for me, which I can see from my initial post, but I'm not so sure.
If it is active by default, then it's reasonable enough to expect the users to know what their options are when they deactivate it. So it's fine as it is and we can close this subject here.
It if isn't, then it's a different story. You say it's the same thing by removing/disabling all the configuration from the normal UI, and that's kind of true, but for someone who needed to do that quickly, they might have to disable Mac Mouse Fix entirely, because they did not know of the existence of that option in the menubar in the first place.

On scrollbars

That was quick! Thanks!

On LinearMouse compatibility and and linear scrolling

There are two things that I experienced with LinearMouse and I couldn't achieve in Mac Mouse Fix, and I'm not sure if they are related.

First: scrolling acceleration. I haven't found a way to disable this in Mac Mouse Fix. I find it distracting when I try to scroll more than 4 times and it starts jumping a lot further and the speeds increases even more from that point on.
I don't have a mouse with unlocked scroll, I just scroll kind of the same amount in a direction, and about the 5th time I do that I start to see an increase in movement, although I didn't change my scrolling pace.

Second: scrolling speed. I've seen this through remote desktop to Windows: I have a notepad open and I start to tick the scrollwheel a couple of times (random amout, depending on what Mac Mouse Fix options I have enabled) to see the text scroll. No matter what options I choose I can never see it scroll immediately.
Scroll Speed: High, Always-On Precise Scrolling: Off: ~2-3 ticks
Scroll Speed: Medium, Always-On Precise Scrolling: Off: ~3-4 ticks
Scroll Speed: High, Always-On Precise Scrolling: On: ~9-12 ticks
My guess is that the acceleration has something to do with it, but also the scrolling speed seems to be higher with LinearMouse and also gives a lot more options on this part. The default (by lines: 3), is about 3 times faster than High with Precise Scrolling off, and goes to a max of 10.

In LinearMouse I can get both of them with: Enable linear scrolling enabled. Through remote desktop in Windows, if I match the "By lines" option in LinearMouse to the "number of lines to scroll at a time" in Windows (see screenshot below), then each tick of the scroll in notepad is an 1 to 1 input with Windows. There might be the odd moment when the input skips the number of lines by 1 or 2, when I scroll fast, but it's hard to notice.

As a conclusion: having an option to disable scrolling acceleration, and adding more scrolling speed options might do it. But this is just a guess based on my perception. It's really difficult for me to say for sure what LinearMouse actually does.

On the Scrolling option in the menubar

I see more value in having it disable everything related to scrolling. You never know the intent of the user, and if they want part of it, they can just let that part enabled in normal UI. Currently there's no way we can get just the button mapping, if that's what we found useful from this app, and to let any other app take care of the scrolling part (it doesn't even have to be LinearMouse, that was just my case).

On the menubar options activating by themselves

I saw that the options in the menubar get active if I just open the related page in the normal UI. Could this be instead something like: keep the part where they both get automatically activated if the menubar is turned off (because there's no option to turn them on), and if the menubar is active then don't change their state.
Having a menubar in which I explicitly went to turn off an option, for me would mean that I need to go back to change it if I don't want that to happen anymore. For example: I could just be comparing the settings with some other app I found - this doesn't mean I wanted it active, and it's worse that there's no indication that this happens.

[krackers]

difficult for me to say for sure what LinearMouse actually does.

From I can piece together (I'm don't know any Swift) I think it operates similar to https://github.com/emreyolcu/discrete-scroll where it effectively uses a cgeventtap to override the scrolling distance like

        int64_t deltaY_raw = CGEventGetIntegerValueField(event, kCGScrollWheelEventPointDeltaAxis1);
        CGEventSetIntegerValueField(event, kCGScrollWheelEventDeltaAxis1, SIGN(deltaY_raw) * LINES);

And the equivalent of windows number of lines per tick would be to modify LINES here.

https://github.com/linearmouse/linearmouse/blob/935d8c1219bab2d8d0cff2f7e99d9a4ad49e777d/LinearMouse/EventTransformer/LinearScrolling.swift#L37-L47