[OpenIndiana-discuss] Fwd: [oi-dev] VirtualBox-5.1.6 on self-built Qt5.5 bins uploaded (built on Hipster 2015 illumos-f83b46b)

Thu Sep 22 03:34:03 UTC 2016

-------- Пересылаемое сообщение --------
От кого: Мартин Бохниг via oi-dev <oi-dev at openindiana.org>
Кому: OpenIndiana Developer mailing list <oi-dev at openindiana.org>
Дата: Четверг, 22 сентября 2016, 3:32 UTC
Тема: [oi-dev] VirtualBox-5.1.6 on self-built Qt5.5 bins uploaded (built on Hipster 2015 illumos-f83b46b)

Hello Aurelien and Hipster community,

as promised I'm absolutely open to using Qt5 rather than Qt4.8, as long as this Qt5.x in question A) builds here on old Hipster 2015 and B) works together with VirtualBox5.x.

The Qt5.5 build process functions much better than 5.8alpha on my 2015 based setup with gcc4.9 and the most important thing: Without -std-gnu++11 (otherwise vbox won't build).
A hack in the middle would be possible (to build Qt5.6++ with -std-gnu++11 (Qt5.6++ *requires* -std-gnu++11)  and run Vbox5.x's ./configure with the same CFLAGS/CXXFLAGS (otherwise Qt5 detection tests FAIL), then killing make/kmk and actually building Vbox5.x without -std-gnu++11.
If that sounds like dirty crap, I agree with you. Also there are 100 more problems with 5.8alpha on my build machine.

Hence - if we really want Qt5.x for Vbox5.x, then me should stick to Qt5.5.
The build works as said much better, but still quits at various places.
However - enough was built in my tests with some hacks and -k to get Vbox linked against Qt5.5.

BENEFITS of Qt5.5 over Qt4.8:

{A}
Now the only stderr messages when starting /opt/VirtualBox/VirtualBox are:

$ vbox
Qt WARNING: QXcbWindow: Unhandled client message: "_GTK_LOAD_ICONTHEMES"
Qt WARNING: QXcbWindow: Unhandled client message: "_GTK_LOAD_ICONTHEMES"
Qt WARNING: QXcbWindow: Unhandled client message: "_GTK_LOAD_ICONTHEMES"

compare this to the old ones with Qt4.8:

$ /opt/VirtualBox/VirtualBox
Qt WARNING: Object::connect: No such signal QApplication::screenAdded(QScreen *)
Qt WARNING: Object::connect:  (sender name:   'VirtualBox')
Qt WARNING: Object::connect: No such signal QApplication::screenRemoved(QScreen *)
Qt WARNING: Object::connect:  (sender name:   'VirtualBox')
Qt WARNING: Object::disconnect: No such signal QApplication::screenAdded(QScreen *)
Qt WARNING: Object::disconnect:  (sender name:   'VirtualBox')
Qt WARNING: Object::disconnect: No such signal QApplication::screenRemoved(QScreen *)
Qt WARNING: Object::disconnect:  (sender name:   'VirtualBox')

{B}
Now you can run /opt/VirtualBox/VirtualBox as root or as su to root without Christian Meier's X11 tunnelling trick.

You need to run Vbox as root for USB device detection and also for Non-USB optical media passthrough (with some MS-Win software needed to pass the security checks required for installation, while reading out hidden data from physical (!!!) optical media.
See  https://openindiana.org/pipermail/openindiana-discuss/2016-September/019561.html
suid doesn't work, nor would it help to make the corresponding devices user-writable.
It is a known Vbox mystery and that's not my thing to get this solved. If you run it as root, then it works. And the good news is, that now you can run also Vbox5.x as root again without the X11 forwarding/tunneling trick.
BTW: Oracle's official Vbox5.0.x and 5.1.x packages had the same off issue as root (complete hang of the Vbox GUI, no text readable).
Now a thing of the past :)

{C}
It looks brighter and smoother now with Qt5.5

{D}
Aurelien and all Qt5.x enthusiasts are (hopefully) happy now.

Here are the new bins (not sure if this self-compiled Qt5.5 works as well for you as the previous test-package did) :

@@@Screenshots:

http://opensxce.org/virtualbox5_qt5.5/testbins_perhaps_instable/Screenshots/Screenshot-163.png
http://opensxce.org/virtualbox5_qt5.5/testbins_perhaps_instable/Screenshots/Screenshot-164.png
http://opensxce.org/virtualbox5_qt5.5/testbins_perhaps_instable/Screenshots/Screenshot-165.png    (with USB devices when running as root)

@@@Updated Install instructions:

http://opensxce.org/virtualbox5_qt5.5/testbins_perhaps_instable/INSTALL.txt

The new part is mostly: 

-0.5) Install instable Qt5.5 64bit /usr/lib/qt/5.5

$ su

# cd /usr/lib/qt/.

# wget wget  http://opensxce.org/virtualbox5_qt5.5/testbins_perhaps_instable/usr_lib_qt_5.5__64BITbins.tar.bz2

# bzcat ./usr_lib_qt_5.5__64BITbins.tar.bz2|tar xvf -

# rm usr_lib_qt_5.5__64BITbins.tar.bz2

@@@Qt5.5 64bit:

(to be extracted in /usr/lib/qt/.)
http://opensxce.org/virtualbox5_qt5.5/testbins_perhaps_instable/usr_lib_qt_5.5__64BITbins.tar.bz2

@@@ New SVR4 Vbox package:

http://opensxce.org/virtualbox5_qt5.5/testbins_perhaps_instable/VirtualBox-5.1.6-SunOS-amd64-r110634.pkg

This Qt5.5 build is still instable and perhaps won't work everywhere (mind the dependencies, and watch out if you are missing any bits)
I share it with you in the hope that some may find it useful (as a preliminary preview).

Because getting Qt5.5 more stable will consume another 1 or 2 days on this slow Celeron G530.

That's it for today,
happy testing and regards,

%martin

-- 
Мартин Бохниг
_______________________________________________
oi-dev mailing list
oi-dev at openindiana.org
https://openindiana.org/mailman/listinfo/oi-dev

----------------------------------------------------------------------