> In conclusion, use (double) instead of (double?).
I'd happily use (double), as it makes sense. But the generated c-code
is wrong and the C-compiler complains:
error: pointer value used where a floating point value was expected
That is quite understandable when looking at the generated c-code,
which tries to cast the incoming pointer generic type directly to
gdouble.
Another example:
public class Test<T> {
public T number;
public void set_number (T number) {
double tmp = (double) number;
this.number = tmp;
}
}
One would assume, that the first line in set_number method is correct
(if the object constructed by "new Test<double>()"). Vala doesn't
complain, but the c-compiler does. The generated C code in this case:
void test_set_number_generic (Test* self, gconstpointer x) {
gconstpointer _tmp0_;
gdouble tmp;
gpointer _tmp1_;
g_return_if_fail (self != NULL);
_tmp0_ = x;
tmp = (gdouble) _tmp0_;
_tmp1_ = ((tmp != NULL) && (self->priv->t_dup_func != NULL)) ?
self->priv->t_dup_func ((gpointer) tmp) : ((gpointer) tmp);
((self->number == NULL) || (self->priv->t_destroy_func ==
NULL)) ? NULL : (self->number = (self->priv->t_destroy_func
(self->number), NULL));
self->number = _tmp1_;
}
Again a pointer _tmp0_ is being casted to gdouble, which is wrong.
When using (double?) instead of (double), the c-code in this line is
like this:
tmp = *((gdouble*) _tmp0_);
This is correct, since _tmp0_ in this case is a pointer to a boolean and
tmp should hold it's value. So quite oddly casting to (double?) seems to
work.
However, the real problem is when trying to assign the value to the
class member:
error: invalid operands to binary != (have ‘gdouble’ and ‘void *’)
error: cannot convert to a pointer type
error: cannot convert to a pointer type
the corresponding line from the c-code above:
_tmp1_ = ((tmp != NULL) && (self->priv->t_dup_func != NULL)) ?
self->priv->t_dup_func ((gpointer) tmp) : ((gpointer) tmp);
So what I need is a way to tell the vala compiler to convert the double
type to the generic type (effectively creating a pointer gdouble*),
which then can be assigned to the classmember (I did this in the
previous email by using the address-of operator, but there has to be a
more vala-way of doing it).
Note: when assigning the classmember from outside the class, everything
is fine:
var test = new Test<double>();
test.number = 1.47;
But how can I do this from inside the class?
Regards,
Stephan
_______________________________________________
vala-list mailing list
vala-list@gnome.org
http://mail.gnome.org/mailman/listinfo/vala-list
