Tue 02 Jun 2015 04:52:50 PM UTC, comment #9:
Retagging release from 4.0.0-rc4 to 4.0.0.
|
Sat 02 May 2015 04:41:30 AM UTC, comment #8:
Oops. Wrong report.
|
Sat 02 May 2015 04:40:50 AM UTC, comment #7:
I checked in a cset that at checks for a modern version of sndfile before setting HAVE_SNDFILE (http://hg.savannah.gnu.org/hgweb/octave/rev/3b3579ad7e46). More could be done, but this is probably enough.
|
Sat 02 May 2015 03:44:29 AM UTC, comment #6:
I unrolled the for loop because it was so small and because I felt it was clearer about what was happening. I checked in the change under your name here (http://hg.savannah.gnu.org/hgweb/octave/rev/6446649e42c1). Closing report.
|
Fri 01 May 2015 02:41:04 PM UTC, comment #5:
I suspect (someone correct if I'm wrong) that the only functions that matter are the ones that pass a pointer or a reference to a qreal, not by value or return value.
There are only a handful of functions in the Qt header files that do so, a quick grep in the Qt directory shows functions like getRect, getCoords, getRgbF, getHsvF, etc.
And the only one of these I see in Octave is getRgbF, so this looks pretty safe and complete to me.
|
Fri 01 May 2015 02:05:13 AM UTC, comment #4:
This appears to be the only compile error on arm. Final patch:
|
Thu 30 Apr 2015 10:12:04 PM UTC, comment #3:
That fails with:
I think it would be:
Ugly, yes, but I don't have another suggestion.
|
Thu 30 Apr 2015 09:58:27 PM UTC, comment #2:
Is it only one location that this happens? Or is the assumption that qreal = double all over the Qt code? If it's just one or two locations then we can probably hack something together as suggested in comment #1.
|
Thu 30 Apr 2015 08:39:45 PM UTC, comment #1:
http://koji.fedoraproject.org/koji/getfile?taskID=9608836&name=build.log&offset=-4000
http://doc.qt.io/qt-4.8/qtglobal.html#qreal-typedef
says:
typedef qreal
Typedef for double on all platforms except for those using CPUs with ARM architectures. On ARM-based platforms, qreal is a typedef for float for performance reasons.
This causes the problem in
I think this is a ugly patch but just to give an idea:
|
Thu 30 Apr 2015 08:08:27 PM UTC, original submission:
graphics/QtHandlesUtils.cc: In function 'Matrix QtHandles::Utils::toRgb(const QColor&)':
graphics/QtHandlesUtils.cc:155:43: error: no matching function for call to 'QColor::getRgbF(double&, double, double*) const'
c.getRgbF (rgbData, rgbData+1, rgbData+2);
^
In file included from /usr/include/QtGui/qpalette.h:46:0,
from /usr/include/QtGui/qwidget.h:50,
from /usr/include/QtGui/QWidget:1,
from graphics/Container.h:26,
from graphics/QtHandlesUtils.cc:37:
/usr/include/QtGui/qcolor.h:114:10: note: candidate: void QColor::getRgbF(qreal, qreal, qreal, qreal) const
void getRgbF(qreal r, qreal g, qreal b, qreal a = 0) const;
^
/usr/include/QtGui/qcolor.h:114:10: note: no known conversion for argument 1 from 'double' to 'qreal {aka float*}'
Makefile:3422: recipe for target 'graphics/graphics_libgui_graphics_la-QtHandlesUtils.lo' failed
|