78450 - Plugin windows don't work with mouse wheel.

Status: RESOLVED DUPLICATE of bug 59211

(Core :: XUL, defect)

Description

[Mark Hammond [:markh] [:mhammond]]

Plugin windows have no opportunity to deal with mouse wheel messages.

The WM_MOUSEWHEEL message is never directed at a plugin window, but always at 
its parent.  The WM_MOUSEWHEEL code in nsWindow.cpp detects that the target of 
the message is not an nsWindow, and assuming the window is in another 
application, simply discards it.

Attaching a proposed patch.  Like bug 75864, it determines if the target window 
has an nsWindow parent, and if it does, forwards the message directly to the 
child.  If the target window is truly outside the app, it is still ignored.

Assigning to Rod and adding av to CC, as it is very similar to bug 75864.

Comment 1

[Mark Hammond [:markh] [:mhammond]]

Attachment: Proposed patch

Comment 2

[Mark Hammond [:markh] [:mhammond]]

I should mention that the patch also removes an unused and non-existing global 
variable g_hinst.  My first attempt at a patch tried to use this, just to find 
it does not exist anywhere and will give link errors if referenced.  This is 
obviously not a critical part of the patch, but is just for cleanup.

Comment 3

[John Morrison]

did someone mention 'mousewheel'? (cc: bryner)

Comment 4

[doctor__j]

Dupe of 59211?

Comment 5

[Mark Hammond [:markh] [:mhammond]]

Agreed - this is a dupe of bug 59211 - however, this bug has a patch, and the 
other doesn't.  Not sure what the protocol is here, so leaving to someone else 
to actually mark one of the bugs as a dupe.

Comment 6

[John Morrison]

I'd go with the one that has working code (but I'll defer to bryner to close 
out the other one). Is this patch ready to go? Does it just need some r/sr 
love?

Comment 7

[Mark Hammond [:markh] [:mhammond]]

Afraid it does need r/sr love.  I don't even know who to approach.

That said, it is still ready to go :)  ActiveState's Komodo has been running 
this for some time.  Not that that is a _great_ indication, but it is not a bad 
one :)

Comment 8

[John Morrison]

I think av and rods are away (or soon to be away), but I think bryner could 
review this. Not sure who the best person would be for sr=, though.

bryner, can you have a look? 

[by the way, the 'g_hinst' is just additional dead code removal, right?].

Comment 9

[Mark Hammond [:markh] [:mhammond]]

'g_hinst' is indeed dead code removal.  Any attempt to use it results in link 
errors.

Comment 10

[rods (gone)]

Yes, this looks ok, but bryner should review it also. (just a nit, put :: in 
front of the SendMessage)

Have hyatt do the super-review.

r=rods

Comment 11

[Mark Hammond [:markh] [:mhammond]]

Attachment: Updated patch with ::SendMessage() as requested.

Comment 12

[Mark Hammond [:markh] [:mhammond]]

Attachment: doh - Identical patch, but this time with -u

Comment 13

[Brian Ryner (not reading)]

r=bryner

Comment 14

[Mark Hammond [:markh] [:mhammond]]

*sigh* - the attached patch is no good.  It will only work with plugins 
designed to handle the mouse-wheel.  It crashes both the sample plugin and the 
flash plugin.

Comment 15

[Mark Hammond [:markh] [:mhammond]]

*** This bug has been marked as a duplicate of 59211 ***