508382 - Wrong property information from plugin “offsetTop” and window object “pageYOffset” returned for windowles content after scrolling

Status: RESOLVED INCOMPLETE

(Core Graveyard :: Plug-ins, defect)

Description

[Andy Rivas]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.1.2) Gecko/20090729 Firefox/3.5.2 (.NET CLR 3.5.30729)
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.1.2) Gecko/20090729 Firefox/3.5.2 (.NET CLR 3.5.30729)

In case of FF windowless mode, we try to get the plugin “offsetTop” and window object “pageYOffset” from NP APIs. So plugin “offsetTop” and window object’s “pageYOffset” is always zero even though the content is scrolled. This mean we couldn’t get the absolute position from the passing relative position.


Reproducible: Always

Steps to Reproduce:
- With latest Firefox, navigate to MouseOffsetBugTestPage.html found in the attached zip
- verify visually that the button responds to enter, leave, down, up correctly
- scroll the browser content a little bit

Actual Results:  
- notice that the button no longer responds to mouse, instead the mouse events fire when you hover or click in the area where the button used to be

Comment 1

[Andy Rivas]

Attachment: Offset bug Sample Application

Comment 2

[Andy Rivas]

Attachment: Offset bug Sample Code

Comment 3

[Andy Rivas]

Can someone provide comment or a status for this bug?  This forum post describes the issue a little more http://www.codeplex.com/deepearth/Thread/View.aspx?ThreadId=47329

Comment 4

[Andy Rivas]

Pinging again for status.

Comment 5

[Johnny Stenback (:jst)]

Roc, is this a regression? If not, should this still block?

Comment 6

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

I don't know if it's a regression. If it is, it's probably a regression in offsetLeft/offsetTop, which are woefully underspecified and fragile.

Andy: thanks for the sample code. I don't know why your code does this, though. These APIs are really nasty and I suspect if you switched to getBoundingClientRect your problems would go away. But can you describe why you need to do this for mouse events at all? We pass in mouse event coordinates that are relative to the window we gave in SetWindow (that was bug 418740).

Comment 7

[Andy Rivas]

I've got this assigned back to one of our devs for further investigation/explanation on this.

Comment 8

[Andy Rivas]

The problem is that FF3.52 sometimes doesn’t update the scrolled mouse position correctly. 

We are using the mouse event and plugin window position. The problem is that the plugin window position doesn’t update correctly from the scrolling; it is ok sometimes but doesn’t work randomly.

Comment 9

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

OK, so the underlying bug is that we're not calling NPP_SetWindow sometimes when we should (scrolling). That should be filed and fixed.

Andy, is it possible for you guys to produce an HTML-only testcase that shows how pageYOffset and/or offsetTop are going wrong?

Comment 10

[Andy Rivas]

Roc,

Stale, I know, but getting more info on this now:

Even calling NPP_SetWindow to update the plugin position, NPAPI still doesn’t update the position sometimes.
For example, fast mouse scroll etc.

Comment 11

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

I'm not sure what you mean. What do you mean "NPAPI doesn't update the position"?

Comment 12

[Andy Rivas]

One of our developers has been looking at this and here's what we have found.

1. Firefox doesn't call into SetWindow when scrolling down (when scrolling up, it is called consistently) which means we can't use the y-value provide by SetWindow to calculate any offset.

2. window.pageYOffset, window.pageXOffset are not accurate when called by a windowless plug-in, so they cannot be used to work around this problem

I'm attaching an html repro for the pageYOffset bug but the SetWindow bug requiress a plugin.

Comment 13

[Andy Rivas]

Attachment: HTML sample of pageYOffset issue

Comment 14

[Benjamin Smedberg]

Resolving old bugs which are likely not relevant any more, since NPAPI plugins are deprecated.