When should Python 3.3 become the default?

[Scott Kitterman]

On Saturday, October 20, 2012 01:45:22 PM Matthias Klose wrote:
...
> If people do not want to address the pyqt "tooling" issues, these
> should be at least documented in the bug tracker.
...

I spent a fair amount of time last night on this.  First, I had to rebuild 
dbus-python and sip4.  dbus-python turned out to be a no-change rebuild (I 
don't know why I got the configure error the first time), but sip4 needed 
packaging changes to deal with multiarch paths (turned out not to be very hard 
- I have an ugly, but working package in my PPA).  

For PyQt, it's the multiarch paths that are causing me grief as well.  I got 
configure working, but as soon as it starts to compile, it explodes thusly:

make[3]: Entering directory `/home/python-qt4-4.9.3/build-3.2/qpy/QtDBus'
g++ -c -pipe -O2 -fPIC -Wall -W -D_REENTRANT -DQT_WEBKIT -DQT_NO_DEBUG -
DQT_DBUS_LIB -DQT_GUI_LIB -DQT_CORE_LIB -DQT_SHARED -
I/usr/share/qt4/mkspecs/linux-g++ -I. -I/usr/include/qt4/QtCore -
I/usr/include/qt4/QtGui -I/usr/include/qt4/QtDBus -I/usr/include/qt4 -
I/usr/include/python3.3m -I../../QtDBus -I../../../qpy/QtDBus -I. -o 
qpydbus_chimera_helpers.o ../../../qpy/QtDBus/qpydbus_chimera_helpers.cpp
In file included from /usr/include/python3.3m/sip.h:32:0,
                 from ../../QtDBus/sipAPIQtDBus.h:33,
                 from ../../../qpy/QtDBus/qpydbus_chimera_helpers.cpp:26:
/usr/include/python3.3m/Python.h:8:22: fatal error: pyconfig.h: No such file or 
directory
compilation terminated.

If I read that correctly, Python.h (part of python3.3-dev) can't find pyconfig.h 
(part of libpython3.3-dev).  It has moved from python3.2 minimal at 
/usr/include/python3.2mu/pyconfig.h to libpython3.3-dev at (for i386) 
/usr/include/i386-linux-gnu/python3.3m/pyconfig.h.  I don't think this is a 
PyQt issue to fix, so I'd appreciate advice (I checked and my chroot does have 
the file in that location).

I'm not arguing if 3.3 should be supported at the start or not, but my 
experience with a couple of complex packages is that it's the multarch path 
changes that are going to be more of a problem.

Scott K