[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#451046: qt4-x11 makes inconsistent use of floats and doubles

Package: qt4-x11
Version: 4.3.2-1
Severity: important
Tags: patch

When investigating the build failure of ktoon on mips/mipsel/arm
I found it comes down to two problems in qt4-x11:

a) It mis-assumes that non-IRIX MIPS systems generally have no FPU.
   The appended patch fixes that.

b) For ARM and MIPS it defines qreal as a float, which confusingly
   means isDouble() returns a float value. But other parts of the
   package, as well as users of qt4-x11, continue to use double types
   instead of qreal, which leads to data type mismatches, and eventual
   build failures.

Unless the intent is to outlaw the use of "double" or "float" for all
uses of qt4-x11 I regard this feature as inherently broken.

I'm also dubious about the performance gain achieved by this hack, I
CC debian-arm for more insight into this.


--- qt4-x11-4.3.2/src/corelib/global/qglobal.h.old	2007-11-10 18:17:21.000000000 +0000
+++ qt4-x11-4.3.2/src/corelib/global/qglobal.h	2007-11-10 18:18:21.000000000 +0000
@@ -840,7 +840,7 @@ QT_BEGIN_HEADER
 typedef int QNoImplicitBoolCast;
-#if defined(QT_ARCH_ARM) || (defined(QT_ARCH_MIPS) && !defined(Q_OS_IRIX))
+#if defined(QT_ARCH_ARM)
 #define QT_NO_FPU

Reply to: