[Bugs] #4132 Write URGE: Xorg pointer emulation issues

[Sugar Labs Bugs]

#4132: Xorg pointer emulation issues
-------------------------+--------------------------------------------------
    Reporter:  garnacho  |          Owner:  garnacho
        Type:  defect    |         Status:  assigned
    Priority:  Urgent    |      Milestone:  0.98    
   Component:  Write     |        Version:  0.97.x  
    Severity:  Critical  |       Keywords:  upstream
Distribution:  OLPC      |   Status_field:  Assigned
-------------------------+--------------------------------------------------

Comment(by dsd):

 I can reproduce the Write issue also in that new version
 (1.13.0-7.fc18.olpc2). The test case is something like this:
  * Start Sugar
  * Do some touchy stuff - open/close the frame
  * Open Write
  * Type in a few lines of text
  * Create a text selection with the touchscreen
  * Use the touchscreen to drag and move the selected text to another
 position in the document. The text should still be selected at the end.
  * Move the mouse. Without clicking or touching, the text will act as if
 it is being dragged by the mouse cursor.
  * Close write with the mouse
  * Open Paint with the mouse
  * Move the mouse cursor around the screen. A line will be drawn as if the
 mouse button is being held down.

 '''xorg-x11-server-1.13.0-7.fc18''' includes: Xi: Call UpdateDeviceState()
 after the first emulated motion event

 This version is working well, w.r.t. this issue. But that patch didn't go
 upstream.

 '''xorg-x11-server-1.13.0-7.fc18.olpc1'''
 drops: Xi: Call UpdateDeviceState() after the first emulated motion event
 adds: Xi: Set modifier mask on touch events
 adds: Xi: Update the device after delivering the emulated pointer
 event(#56558)

 The non-upstream patch was dropped, and was replaced by the 2 upstream
 patches which are supposed to be equivalent.

 This introduced the bug described here.

 '''xorg-x11-server-1.13.0-7.fc18.olpc1''' adds Xi: update device state
 even if no emulated events are sent

 The bug described here is still present.

 ----

 So, I think for the time being, we can just downgrade to
 xorg-x11-server-1.13.0-7.fc18, going back to the non-upstream " Xi: Call
 UpdateDeviceState() after the first emulated motion event" approach. I
 believe the only reason we started shuffling the patches around in .olpc1
 (causing breakage) was to adopt the way that upstream fixed the problem.
 But the upstream fixes now don't seem to be equivalent, and I don't think
 adopting those patches fixes any other issues we were seeing. Am I missing
 anything?

 Obviously we do want this fixed properly and fixed upstream, but just in
 terms of restoring the non-buggy behaviour we had a couple of weeks ago,
 downgrading this package for the time being could make sense for us.

-- 
Ticket URL: <http://bugs.sugarlabs.org/ticket/4132#comment:14>
Sugar Labs <http://sugarlabs.org/>
Sugar Labs bug tracking system