allow scrolling past child layer, when a new scroll starts at the end of child layer

[gterzian]

WIP for servo/servo#13249

---

This change is

[bors-servo]

☔ The latest upstream changes (presumably #488) made this pull request unmergeable. Please resolve the merge conflicts.

[glennw]

r? @pcwalton

[gterzian]

To summarize what was discussed in servo/servo#13249

• "If a new scroll gesture starts while you're already scrolled all the way to the edge of the scrollable element under the mouse, then the scroll will be handed off to the next outer scrollable element which is actually scrollable in that direction" servo/servo#13249 (comment)
• "If you scroll a long page, and somewhere on that page is a smaller scrollable element, you don't want that element to swallow the rest of your scroll gesture if it suddenly happens to move under your mouse. You want the current scroll gesture to keep scrolling the outer page." servo/servo#13249 (comment)

[pcwalton]

Is this ready for review and ready to go?

[gterzian]

@pcwalton yes it is, thank you for your time.

[gterzian]

@pcwalton since from the issue itself it might not be clear what the goal of this PR is, there are two:

1. When a child layer has scrolled to it's end, and a new scroll gesture is started, start scrolling the nearest parent layer. Current behavior: one is stuck inside the child layer.
2. When one is scrolling a layer, and one encounters a scrollable child layer, one should keep scrolling the currently scrolling parent layer. Current behavior: the scroll is 'captured' by the child layer, which starts scrolling.

See the comments above for links to the discussions around this within the issue.

[bors-servo]

☔ The latest upstream changes (presumably #517) made this pull request unmergeable. Please resolve the merge conflicts.

[gterzian]

@pcwalton @glennw Since there was quite a big change in the scroll function, I am currently looking if my changes are necessary at all, and if so, how to re-implement based on the latest... I'll let you know when that is done...

[mrobinson]

@gterzian Let me know if you need my input for that. I'm sorry that my change stomped on yours!

[gterzian]

@mrobinson thanks for your offer of input, and no problem at all! 😄

Let me get a better look at the changes, and test the latest behaviour, then I'll get back to you...

[gterzian]

@pcwalton @glennw ok from my initial testing it seems that the behavior suggested in #465 (comment) is still something that needs to be implemented.

@mrobinson perhaps it's best if I wait until the servo part of you work has been merged, and then use that as a basis for exploring how to re-implement my changes?

[gterzian]

closing, in favor of #533