On 2010-09-06 10:44, Jacob Carlborg wrote:
On 2010-09-06 05:15, dsimcha wrote:
I've started playing around with Orange a little to see whether it
would meet
D's cloning needs. IMHO one must-have feature for proper cloning that
truly
"just works" is full aliasing preservation. For example, the following
code
modified slightly from the Orange example doesn't work properly:
import orange._; // import the whole library
class A
{
int[] arr1;
int[] arr2;
equals_t opEquals (Object other)
{
if (auto a = cast(A) other)
return a.arr1 == this.arr1&& a.arr2 == this.arr2;
return false;
}
}
void main ()
{
auto a = new A; // create something to serialize
a.arr1 = [1,2,3,4,5];
a.arr2 = a.arr1[1..$];
auto serializer = new Serializer!(XMLArchive!());
auto data = serializer.serialize(a);
println(data);
auto a2 = serializer.deserialize!(A)(data);
assert(a == a2);
a2.arr2[0] = 0;
println(a2.arr1); // [1,2,3,4,5]
}
Note that Orange gets this right for class references that point to
the same
object, but not for arrays that overlap.
A few questions:
1. Are most serialization libraries in other languages capable of getting
this right?
2. Do others agree that full aliasing preservation is essential with
regard
to array slices?
3. Jacob, do you think you could fix Orange to support this properly?
I can have a look and see what I can do. Could you please report a
ticket so I don't forget.
I've looked at this problem now but I don't know I can detect the
aliasing. Suggestions ?
--
/Jacob Carlborg
