#4881 HIGH Never A: Can't start sugar in remote X.

[Zarro Boogs per Child]

#4881: Can't start sugar in remote X.
------------------------------+---------------------------------------------
 Reporter:  MitchellNCharity  |       Owner:  marco         
     Type:  defect            |      Status:  new           
 Priority:  high              |   Milestone:  Never Assigned
Component:  sugar             |     Version:                
 Keywords:                    |    Verified:  0             
------------------------------+---------------------------------------------
 The ability to run sugar in a remote X server was lost in build 603.

 The recipe had been init 3, then as olpc with environment variables from
 /usr/bin/sugar, run DISPLAY=whatever sugar.  In 603 this started failing
 with
  AttributeError: 'HardwareManager' object has no attribute '_master'
 The current 625, and all the builds I've tested in between, fail
 similarly.

 The apparently same problem can be replicated without setting up for
 remote X.  On both emulation, and a B4 running 623,
  As root, /sbin/init 3.
  su olpc
  grep export /usr/bin/sugar > hee
  source hee
  (sleep 10; X)&
  DISPLAY=:0 sugar
  ctrl-alt-erase to kill X, and see the error message.

 The full traceback follows.  The CRITICAL and XShm messages have been
 occuring "forever", without apparent difficulty.

  ** (sugar-shell:1461): CRITICAL **: _wrap_gst_mixer_list_tracks:
 assertion `GST_IS_MIXER (self->obj)' failed
  mbpixbuf: unable to use XShm. DISPLAY remote?
  Traceback (most recent call last):
    File "/usr/bin/sugar-shell", line 144, in <module>
      hw_manager.startup()
    File "/usr/share/sugar/shell/hardware/hardwaremanager.py", line 88, in
 startup
       self.set_volume(profile.sound_volume)
    File "/usr/share/sugar/shell/hardware/hardwaremanager.py", line 65, in
 set_volume
      if not self._mixer or not self._master:
  AttributeError: 'HardwareManager' object has no attribute '_master'

 Simply wrapping the first use of _master in get/set_volume with a try, and
 returning upon exception, permits sugar to boot.  Though seems not a real
 fix.

 Remote X is currently the only known way to get 1200x900 in emulation.
 Aside from development, automated testing (one of my next tasks), probably
 has to run on emulation, rather than on sugar-jhbuild or real hardware.
 Thus the Priority: high.

-- 
Ticket URL: <http://dev.laptop.org/ticket/4881>
One Laptop Per Child <http://dev.laptop.org>
OLPC bug tracking system