[oi-dev] QT5.8 alpha1 testing build

Мартин Бохниг opensxce at mail.ru
Wed Sep 21 09:29:33 UTC 2016


Hello Aurélien,

thanks for your understanding and for your qualified response.
Well, from vbox's viewpoint Qt.n (with n ==4.x or 5.x or z element N) is just a single of many related aspects.
It just has to perform its job.

On the other hand I certainly understand that Qt by itself is an amazing framework and that it can be someone's pet project in its own sense.
Certainly do ship Qt5 asap (and I give you back the hipster integration 64bit changes of your original diffs, just for having a look), software packages depending on it can slowly get used to it. And then at some future point decide to move over.
In my view it should perhaps be started with 5.5 because it is the last release that build without -std=c++11 (or compatible).
While the later releases not only require that, but also force its clients to be compiled the same way (via their default config scripts anyway).

For another round I'm now trying to test Qt5.5 for a last time together with vbox:


/code/ALP__KMS_MATE__checkout20160911sun/oi-userland/components/qt5$ gmake clobber;gmake build
/bin/rm -f -r qt-everywhere-opensource-src-5.5.0 required-pkgs.mk /code/ALP__KMS_MATE__checkout20160911sun/oi-userland/components/qt5/build/mkgeneric-perl /code/ALP__KMS_MATE__checkout20160911sun/oi-userland/components/qt5/build/mkgeneric-python /code/ALP__KMS_MATE__checkout20160911sun/oi-userland/components/qt5/build/mkgeneric-ruby
/bin/rm -f -r /code/ALP__KMS_MATE__checkout20160911sun/oi-userland/components/qt5/build /code/ALP__KMS_MATE__checkout20160911sun/oi-userland/components/qt5/build/prototype/i386
/bin/rm -f -r qt-everywhere-opensource-src-5.5.0.tar.gz
/bin/rm -f -r /code/ALP__KMS_MATE__checkout20160911sun/oi-userland/components/qt5/qt-everywhere-opensource-src-5.5.0
/PHOTOSVIDEOS1000GB/COMPILE/00__ILLUMOS/code20160608wed/04__solaris11/code/ALP__KMS_MATE__checkout20160911sun/oi-userland/tools/userland-unpack  /PHOTOSVIDEOS1000GB/COMPILE/00__ILLUMOS/code20160608wed/04__solaris11/code/ALP__KMS_MATE__checkout20160911sun/oi-userland/archives/qt-everywhere-opensource-src-5.5.0.tar.gz


But as said: Vbox has different complexities to consider.
Before release (rather than in a future upgrade) it is actually my goal to get these server oriented features compiled in, too:

  --enable-webservice      enable the webservice stuff
  --enable-vnc             enable the VNC server

And this intruduced additional dependencies, which also needed to get added: gsoap and libvnc.
So as you see, Qt4 works(T.M.) plus exists and is available for older 2015 Hipster and I move on (except for a last Qt5.5 test which I will cancel unless it works smoothly in 24 hours).


And then of course there are all the other parts of my Hipster TODO list as mentioned 2 days ago.
Let completely alone that I now finally have to quit university (without Maths degree) and need to register as unemployed.
So - qt5 has exacvtly 24 hours or vbox5 takes 4.8 for now. It is only one small item in a huge puzzle. Things need to move on here  ...



Kind regards,
%martin









>Среда, 21 сентября 2016, 8:44 UTC от Aurélien Larcher <aurelien.larcher at gmail.com>:
>
>Hello Martin,
>
>> I'm by NO means intending to unintentionally diminish your qt5 work.
>> All I'm saying is: Qt5.8 is simply too new in every aspect.
>> It needs to be migrated in carefully and in paralel to good old stable 4.8
>
>No offense taken ;)
>I agree with you that we should stick with Qt4 for now and that I will
>provide Qt5 alongside (in /usr/bin/qt/5.8) as experimental package for
>testing.
>
>>
>> As promised I now wanted to port vbox5 back to Qt5 and run a test compile
>> against your bins.
>
>I've spun a testing build of VirtualBox last week end:
>- one target calling dtrace with no probe had to be removed from a Makefile,
>- vboxusb does not work of course since I did not apply your fixes,
>- it can only be run as root due to a plugin load permission issue in Qt,
>- the keyboard input does not work for some input widgets.
>
>I could boot a Firefly image in it.
>
>Anyway you are correct that we should do not waste more time on Qt5
>since you proved that Qt4 offers better support right now.
>We will see next year if/when Qt5 is required and in better shape on
>OI (hopefully because we upstreamed patches :D).
>
>Kind regards
>
>Aurelien

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://openindiana.org/pipermail/oi-dev/attachments/20160921/864ce778/attachment-0005.html>


More information about the oi-dev mailing list