1536520 - Mouse wheel input interrupts autoscroll (middle click scroll)

Status: NEW

(Core :: Panning and Zooming, defect, P3)

Description

[Mario Haner]

User Agent: Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:65.0) Gecko/20100101 Firefox/65.0

Steps to reproduce:

1. Use autoscroll on any page (hold middle mouse button down and move the cursor).
2. Slightly nudge the mouse wheel up or down at the same time.

Actual results:

Auto scrolling is interrupted.

Expected results:

Auto scrolling should have continued since the middle mouse button was not released.

Comment 1

[Mario Haner]

You may have to activate autoscrolling for this:

Preferences -> General -> Browsing -> Use autoscrolling

Comment 2

[Mario Haner]

I used this page[1] to verify that there is in fact no MouseUp event to rule out a broken mouse button etc.

[1] Mouse Event Test Page: https://unixpapa.com/js/testmouse.html

Comment 3

[Kartikaya Gupta (email:[e-mail])]

This sounds like the expected behaviour to me. Why do you think auto scrolling should continue after a wheel event?

Comment 4

[Kartikaya Gupta (email:[e-mail])]

Hm, actually after some testing (on Windows anyway) it does look like other apps don't interrupt autoscroll with wheel events. I could have sworn they used to do this.

Comment 5

[Mario Haner]

(In reply to Kartikaya Gupta (email:[e-mail]) from comment #3)

This sounds like the expected behaviour to me. Why do you think auto scrolling should continue after a wheel event?

(In reply to Kartikaya Gupta (email:[e-mail]) from comment #4)

Hm, actually after some testing (on Windows anyway) it does look like other apps don't interrupt autoscroll with wheel events. I could have sworn they used to do this.

Some mouses have very sensitive wheels and it can cause scrolling to be interrupted unintentionally. I believe staying in autoscroll mode provides better usability here and other browser developers seem to have come to the same conclusion.

Comment 6

[Botond Ballo [:botond]]

(In reply to Kartikaya Gupta (email:[e-mail]) from comment #4)

it does look like other apps don't interrupt autoscroll with wheel events.

Other apps? I thought autoscroll was a Firefox-specific thing.

Comment 7

[Mario Haner]

(In reply to Botond Ballo [:botond] from comment #6)

Other apps? I thought autoscroll was a Firefox-specific thing.

On Windows, I think every browser has it. On Linux, Firefox is the only browser with native autoscroll. There are extensions for Webkit-based browsers.

Comment 8

[Kartikaya Gupta (email:[e-mail])]

(In reply to Botond Ballo [:botond] from comment #6)

Other apps? I thought autoscroll was a Firefox-specific thing.

I thought it was a standard Windows feature and things like MS Word etc had it too. But maybe I'm misremebering. I've definitely used it in other browsers (old Opera specifically) though and when I wrote comment 4 I had tested in Chrome and Edge on Windows.

Comment 9

[see]

Bug still persists till today after 4 years.
User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:109.0) Gecko/20100101 Firefox/109.0

I had been suffering from my daily usage of Firefox a lot by this bug ever since I switched back from Chrome and Edge more than a year ago. I've been looking for ways to fix this on my own but failed.

<rant>
I'm actually very disappointed by how the Firefox project is being run these years. Not only Firefox has seemingly stopped innovating, bugs like this can actually be ignored for 4 years with no one looking into it to even assess its severity, impact and effort needed to fix. It's no wonder that people are leaving Firefox. If it's not because of the need to support diversity in browser engines, I wouldn't have switched back to Firefox, effectively abandoned my established workflow on Edge with its features like collections and native vertical sidebar, and other extensions that had worked well for me.
</rant>

Comment 10

[Botond Ballo [:botond]]

We receive a higher volume of bug reports than we're able to fix. (To get a sense of the scale, bug numbers are auto-incrementing, so as of the time of filing this bug, there have been over 1.5 million Bugzilla tickets filed. Today, there are over 1.8 million.)

I had been suffering from my daily usage of Firefox a lot by this bug

We do appreciate and take into consideration information like this that helps us understand the impact of a bug. However, it would be even more helpful to explain how this bug has a significant impact on your Firefox usage.

Are you finding that you often accidentally roll the mousewheel while autoscrolling, when you don't intend to?

When I play around with this, I find that I can reliably autoscroll without rolling the mousewheel. When autoscrolling, I middle click the mousewheel (down and back up), which causes the autoscroll anchor to appear. I can then move the mouse around and scroll without keeping my finger on the mouse wheel (and therefore, avoid unintentionally rolling it).

If you're doing something different, please elaborate. Perhaps you're using a touchpad?

---

One challenge with bugs like this is that the expected behaviour isn't always obvious. There may well be other users who have been relying on the fact that a mousewheel is one way to interrupt autoscrolling, and would be surprised if that behaviour is changed. (In fact, bug 1477995 is an example report where this is the expected behaviour described.)

Even if we agree the mousewheel shouldn't cancel the autoscroll, we need to decide:

1. Should the scrolling the mousewheel normally causes be ignored?
2. Or should it happen in parallel with the autoscroll?

In the case of (2), we would run into an implementation challenge, at least when smooth scrolling is enabled, because autoscroll and mousewheel smooth scroll are both modelled as scrolling animations, and our currently scrolling infrastructure only allows one animation to be running at a given time.

Comment 11

[see]

When I play around with this, I find that I can reliably autoscroll without rolling the mousewheel. When autoscrolling, I middle click the mousewheel (down and back up), which causes the autoscroll anchor to appear. I can then move the mouse around and scroll without keeping my finger on the mouse wheel (and therefore, avoid unintentionally rolling it).

If you're doing something different, please elaborate. Perhaps you're using a touchpad?

No, I'm not using a touchpad. It's the same behavior on my desktop computer and laptop with touch pad when I'm using my mouse Logitech M590. I believe it is not specific to the scroll wheel on this mouse. Scrollwheels on modern mice today have very small scrolling steps (the ticking feedback as we scroll the scrollwheel) and it requires very little force to trigger the scrolling. When middle clicking it, it is very easy to also accidentally scroll the wheel at the same time on these modern mice today.

This is not an issue on other browsers (and also on all other applications with autoscrolling feature), when the autoscrolling gesture is active, the scrollwheel regular scrolling detection will be ignored so that it doesn't cancel out the autoscrolling gesture. Only on Firefox that it is still unintuitively being detected that it ends up cancelling the gesture before the user can interact with the page using the gesture.

If you are not able to trigger this, it is likely that your mouse wheel require a higher threshold to trigger its scrolling and therefore it is unlikely that you trigger both regular mousewheel scroll when trying to middle click to activate the autoscrolling gesture.. You can simulate this behavior by middle clicking activate the autoscrolling gesture, then scroll your mouse wheel. The mouse wheelscroll action will override and cancel off the autoscrolling gesture. Imagine this happens on a scroll wheel that is more sensitive. The accidental cancellation of the autoscrolling gesture happens almost instantaneously when we click on the mouse wheel whenever the force was not directed precise enough (in terms of angle and force) to not scroll the wheel at the same time. It takes an extremely conscious and careful effort for us to exert the right amount of force and angle with our finger on the scroll wheel to trigger autoscrolling gesture without cancelling it out instantaneously. This extremely conscious and careful effort required is unnatural, unintuitive and not user-friendly.

One challenge with bugs like this is that the expected behaviour isn't always obvious. There may well be other users who have been relying on the fact that a mousewheel is one way to interrupt autoscrolling, and would be surprised if that behaviour is changed. (In fact, bug 1477995 is an example report where this is the expected behaviour described.)

To decide on the expected behaviour, I suggest we refer to the implementation on other browsers like Chrome, Edge and Safari. I believe Firefox is the only odd one out in this case. If we don't support it the same way, users from other browsers will feel out of place when trying to switch over to Firefox. In fact, if you are able to try this gesture in older versions of Firefox, e.g. 2.0, I believe the same behaviour was exhibited. As far as I remember, this behaviour only started to change when a major version was released several years ago. (And then I got so frustrated with it and decided to switch to Chrome)

Even if we agree the mousewheel shouldn't cancel the autoscroll, we need to decide:

1. Should the scrolling the mousewheel normally causes be ignored?
2. Or should it happen in parallel with the autoscroll?

I still suggest we refer to other browsers for a consistent behaviour. Firefox is pretty much the only odd one out.

If you're still not convinced that that should be the correct behaviour, then you can do some forensic studies with every major older versions of Firefox with the autoscrolling gesture. You will then be able to see how the behaviour was like, when it was changed, and trace the history to find out whether the change was introduced unintentionally.

Comment 12

[see]

We receive a higher volume of bug reports than we're able to fix. (To get a sense of the scale, bug numbers are auto-incrementing, so as of the time of filing this bug, there have been over 1.5 million Bugzilla tickets filed. Today, there are over 1.8 million.)

I pity you and your team at grounds trying to support us. But Mozilla management has obviously failed to grow the support/bug fix teams in accordance to the growth of users and bug reports, causing difficulties for the staff to catch up with the tickets. And over the years, Mozilla management has not been taking this matter seriously and allow the situation to continue to deteriorate and jeopardize the users share of Firefox, as what we are seeing today. I hope someone with the authority in Mozilla org can realize this and take the necessary steps to restore the glory of Firefox.

Comment 13

[Botond Ballo [:botond]]

Thanks for the additional info. The fact that the accidental cancellation occurs when initiating the autoscroll is particularly relevant, and I can see how that's easy to trigger with a sensitive mouse.

Would you say that ignoring mousewheel scrolls within a short time window (say, half a second) of autoscroll starting, would substantially address the issue as you experience it?

That would be a straightforward fix to implement, and would avoid some of the problems I mentioned above:

• Users who may use mousewheel to intentionally cancel autoscroll later on can continue doing so
• We don't have to deal with performing the mousewheel scroll in parallel with the autoscroll (if the mousewheel happens within that half-second window, it was almost certainly accidental, and can just be ignored)

Comment 14

[see]

Would you say that ignoring mousewheel scrolls within a short time window (say, half a second) of autoscroll starting, would substantially address the issue as you experience it?

Yes, I think that will be sufficient to address the issue.