1262264 - Improve APZ minimap rendering behaviour for subframes with tall viewports

Status: NEW

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

Description

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

+++ This bug was initially created as a clone of Bug #1259529 +++

Bug 1259529 improved the minimap rendering in subframes somewhat, but on some pages it's still not useful. The specific page I have in mind is http://www.sjsu.edu/openuniversity/schedule/. When drawing the minimap, we use either the transformed aClipRect, or the layer's GetEffectiveClipRect(), if there is one, to bound the minimap. However on this page, the GetEffectiveClipRect() is ginormous, and the aClipRect is the one we want. So really I think we want to intersect the two or something, and use that as the bound of the minimap, so that it fits into the smallest clip applied to the layer. I'll play around with this.

Comment 1

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

There's also a second issue where the minimap for the parent frame gets drawn into the subframe as well, because there are multiple layers with the outer metrics.

Comment 2

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

Attachment: Further restrict the minimap to the intersection of the clips

This fixes the issue in comment 0. I can't think of an easy way to fix the issue in comment 1.

Comment 3

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

I don't think this patch is worth landing without fixing the issue in comment 1, because the minimap is useless either way without that. However I don't have any good solutions in mind and I'm not actively working on this, so unassigning.

Comment 4

[Botond Ballo [:botond]]

The issue discussed in comment 1 has been resolved by the WebRender implementation of the minimap in bug 1604280.

The issue discussed in comment 0 remains valid. While the structure of the page at http://www.sjsu.edu/openuniversity/schedule/ has changed since the time of filing the bug and no longer contains a subframe, the earlier page can be seen at https://web.archive.org/web/20160420064916/http://www.sjsu.edu/openuniversity/schedule/. Basically, it has a subframe whose viewport is taller than the browser window, and the minimap is sized to fit the subframe viewport, which means it is also taller than the browser window and so only part of the minimap is visible at a given time.

A solution to this would involve scaling the minimap to the intersection of the subframe viewport and any enclosing viewports (similar to what the attached patch was trying to do in the Layers implementation).