I’ve found that the Qt 4.4.3 libraries supplied with DigiEL-5.0 have a problem converting between double precision floating point and numerical ASCII representations.
The symptoms that led me to finding this are:
Application locks up using 100% CPU when setting a style sheet that uses a gradient fill pattern.
Application locks up using 100% CPU when using QString to format a double precision number (using QString::arg()).
To fix this problem, the qt-embedded libraries need to be rebuilt, specifying -armfpa on the configure script’s command line. Without -armfpa, the script tries to auto-detect the target’s floating point byte order, but fails and picks an incorrect default, resulting in the libraries using the wrong byte order for floating point internal representations.