1345989 - Regression: mousewheel over autoscroll should scroll 1px

Status: RESOLVED INVALID

(Toolkit :: UI Widgets, defect)

Description

[tustamido]

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:52.0) Gecko/20100101 Firefox/52.0
Build ID: 20170302120751

Steps to reproduce:

- Middleclick on the page to show autoscroll icon;
- With the cursor at the height of the autoscroll icon, turn mousewheel.


Actual results:

The page scrolls as if autoscroll isn't activated.


Expected results:

Should scroll up or down by 1px, as it was until Firefox 51 (at least on Windows). This is the easiest way to scroll an exact amount of pixels.

This was caused by Bug 643770 fix patch.

My suggestion to fix this regression:
https://bugzilla.mozilla.org/show_bug.cgi?id=643770#c5

Comment 1

[YF (Yang)]

I don't see the changes between Fx52 and Fx51, on Win10. It is still one by one pixel scroll, in my view.

Comment 2

[tustamido]

Tested in safemode and is still reproducible.

And the change makes sense if you see the 643770's patch. It prevents the default of scroll event if autoscroll is enabled. But this should happend only if the page is scrolling, that is, if cursor is NOT at the height of autoscroll icon.

Comment 3

[:Gijs (he/him)]

Neil, do you know what's going on here? Moving to the same component as bug 643770.

Comment 4

[Neil Deakin]

You could probably just remove the preventDefault call in the DOMMouseScroll event handler added in 643770.

Comment 5

[:Gijs (he/him)]

(In reply to tustamido from comment #0)
> Should scroll up or down by 1px, as it was until Firefox 51 (at least on
> Windows). This is the easiest way to scroll an exact amount of pixels.

In fact, why is this expected? Doesn't seem to work like this in either Edge or IE on my Win10 machine...

Comment 6

[tustamido]

Because Firefox works like this since *always* and this is not inconvenient, on the contrary, is the easiest way to scroll with accuracy (without having to run JavaScript scrollBy). To my knowledge there has never been a complaint about this behavior, so it doesn't hurt, nor impact performance.

The fix is simple, just prevent Bug 643770 patch when the cursor is at the height of the autoscroll icon:
https://bugzilla.mozilla.org/show_bug.cgi?id=643770#c5

Comment 7

[:Gijs (he/him)]

(In reply to tustamido from comment #6)
> Because Firefox works like this since *always* and this is not inconvenient,
> on the contrary, is the easiest way to scroll with accuracy (without having
> to run JavaScript scrollBy). To my knowledge there has never been a
> complaint about this behavior, so it doesn't hurt, nor impact performance.

bug 387521, bug 213496, maybe bug 319080 (which was closed due to lack of response from the reporter).

I'm unconvinced fixing this bug is a net benefit to most users.

Neil, do you have thoughts on what the actual 'expected' behaviour should be, and what to do here? I'm tempted to WONTFIX.

Comment 8

[tustamido]

Only bug 387521 seems to be related. Bug 213496 is the same as bug 643770 (wants to stop autoscroll by turning wheel) and bug 319080 is the opposite (wants that mousewheel scrolls the page normally, as if autoscroll is not running). Neither of these two   interferes with the regression of this bug.

Comment 9

[Neil Deakin]

It looks like we have the same behaviour as IE so I see no need to change this.

Comment 10

[:Gijs (he/him)]

Marking invalid per the discussion in earlier comments. The current behaviour seems correct when looking at other apps, so in this sense the previous behaviour was a bug that is now fixed, and we won't revert that.