On Jan 21, 2013, at 2:36 AM, Stephen Kelly wrote:
> Yes. But note that the user does not have to do anything special in order to
> put the container in a QVariant.
> Not needed at all.
*That* is what I didn't understand -- and I don't see any notion of that in the
QMetaType docs, but that's okay; I'm just glad for the clarification.
My attempt to directly use Q_DECLARE_METATYPE_TEMPLATE_2ARG failed, and I'm
pretty sure that's because the first template argument is not really a class
but instead is a base data type, qint8.
So, I copied it out and started modifying it (outside of any macro) and ended
up with the following, which seems to work fine:
template< qint8 N, class DTYPE >
struct QMetaTypeId< QNDArray<N, DTYPE> >
{
enum {
Defined = QMetaTypeId2<DTYPE>::Defined
};
static int qt_metatype_id()
{
static QBasicAtomicInt metatype_id = Q_BASIC_ATOMIC_INITIALIZER(0);
if (const int id = metatype_id.loadAcquire())
return id;
QVarLengthArray<char, 24> name;
name.append("QNDArray", sizeof("QNDArray") - 1);
name.append('<');
QByteArray tName = QByteArray::number(N);
name.append(tName.constData(), tName.size());
name.append(',');
const char *uName = QMetaType::typeName(qMetaTypeId<DTYPE>());
Q_ASSERT(uName);
name.append(uName, int(strlen(uName)));
if (name.last() == '>')
name.append(' ');
name.append('>');
const int newId = qRegisterNormalizedMetaType< QNDArray<N, DTYPE> >(
QByteArray(name.constData(), name.size()),
reinterpret_cast< QNDArray<N, DTYPE> *>(quintptr(-1)));
metatype_id.storeRelease(newId);
return newId;
}
};
Where I'm not really happy with it is in the use of a QByteArray, as I think
that the purpose for using QVarLengthArray in the original macro was to avoid
heap allocations such as QByteArray will necessarily bring.
Is there something that can be done about that?
Thanks!
Glen
_______________________________________________
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development
