414654 - Accessibles not hidden for visibility:hidden w/ position:absolute

Status: RESOLVED FIXED

(Core :: Disability Access APIs, defect)

Description

[Aaron Leventhal]

Attachment: Testcase

When an object has visibility: hidden it normally does not get an accessible created for it.

However, if position: absolute is also used then we are creating an accessible.

This is because we get PR_TRUE returned from
frame->GetStyleVisibility()->IsVisible()

Robert, do you know if that is intentional?

Comment 1

[Aaron Leventhal]

Attachment: We should not have to do this -- layout is lying to us

Comment 2

[Aaron Leventhal]

Attachment: Correct patch. Should not have to do this but layout is lying to us

Comment 3

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

I bet the problem is that the placeholder frame is marked visible but the out of flow frame is not. That actually makes some sort of sense, pretty much all style properties apply to the out of flow frame instead of the placeholder frame, which is logical if you think about it.

So what you should be doing is calling nsPlaceholderFrame::GetRealFrameFor, but since that's static, I guess you'll have to reimplement that code yourself. Then use the returned frame for style checks such as GetStyleVisiblity()->IsVisible()

Comment 4

[Aaron Leventhal]

Comment on attachment 300121 [details] [diff] [review]
Correct patch. Should not have to do this but layout is lying to us

Roc says to add a method to pres shell to get the real frame for a placeholder frame.

Comment 5

[Aaron Leventhal]

Attachment: Add GetRealPrimaryFrameFor() and fix comments so GetPrimaryFrameFor() doesn't say it supplies the real frame

Comment 6

[Aaron Leventhal]

Attachment: Correct one

Comment 7

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

Comment on attachment 300161 [details] [diff] [review]
Correct one

+   * Gets the real primary frame associated with the content object. This is a
+   * helper function that just forwards the request to the frame manager.

Well, no it doesn't. Fix that comment.

Comment 8

[Mike Beltzner [:beltzner, not reading bugmail]]

Comment on attachment 300161 [details] [diff] [review]
Correct one

a=beltzner