380094 - make mousethrough more generally useful

Status: RESOLVED WONTFIX

(Core :: Layout, defect)

Description

[Chris Toshok]

mousethrough would be extremely useful in more cases than it's used now.  Imagine translucent divs which overlay the actual user interface without keeping the user from interacting with that interface.

Comment 1

[Chris Toshok]

Attachment: first pass

Move the mousethrough handling up to nsFrame from ns{Leaf}BoxFrame, and compute it when we need it instead of updating it only when the attribute is changed (and stuffing the result in an instance variable).

Comment 2

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Comment on attachment 264171 [details] [diff] [review]
first pass

+  virtual PRBool GetMouseThrough() const;

This does not need to be virtual.

Post a new version of the patch with that change, then mark this [checkin needed] and get someone to land it.

Comment 3

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Actually, this is slightly annoying because it means if you set mousethrough on the root element then event targeting becomes O(N^2) in the depth of the frame tree. But I don't think that's worth worrying about.

Comment 4

[Chris Toshok]

Attachment: round two

made nsFrame::GetMouseThrough non-virtual.

Comment 5

[Nickolay_Ponomarev]

Out of interest, can you describe in plain words what this patch changes for the developer? Is this something that needs changes in the documentation? Can this be tested automatically via mochitest?

Comment 6

[Nickolay_Ponomarev]

Checked in, but I'm still interested in the answers to the above questions.

mozilla/layout/generic/nsFrame.cpp              3.730
mozilla/layout/generic/nsFrame.h                3.266
mozilla/layout/xul/base/src/nsBoxFrame.cpp      1.328
mozilla/layout/xul/base/src/nsBoxFrame.h        1.115
mozilla/layout/xul/base/src/nsLeafBoxFrame.cpp  1.61
mozilla/layout/xul/base/src/nsLeafBoxFrame.h    1.29

Comment 7

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Why is this an attribute rather than a CSS property?  I really don't like attribute handling in frames, and adding a proprietary attribute to all languages without any prefixing seems pretty extreme.

Comment 8

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

That said, SVG already defines a CSS property for this:
http://www.w3.org/TR/SVG11/interact.html#PointerEventsProperty

Comment 9

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

(Although the default value for pointer-events doesn't really match the default value we'd want for non-SVG content for compatibility, which is 'visible' rather than 'visiblePainted'.  But we could work around that with a rule in svg.css, I think.)

Comment 10

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

(Actually, we'd have to work around it with a new value (-moz-auto?) since it's inherited.)

Comment 11

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

I filed bug 380573 on that.

Comment 12

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

I think we should back this patch out, since I think adding a new attribute to every element in every namespace is something we should be very hesitant to do.  (As far as I know, this is the first time we've done it, although we discussed doing it for 'id'.)

Comment 13

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

(Well, the first time we've done it for things other than elements with XUL display types.)

Comment 14

[Robert O'Callahan (:roc) (email my personal email if necessary)]

OK sure. I'll back it out when the tree reopens.

Comment 15

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Backed out.

Comment 16

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

So, as I said above (and I think in email as well), I don't think we want to do this -- I'd vastly prefer the approach in bug 380573.  Sorry for all the confusion here.