Issue2451

I'm trying to use ByteArrays in a dict and I get duplicated keys,
It appears that in Jython bytearray can be hashed (not in regular python 2.7) but the hash is not consistent with the equality operator:

>>> from array import array
>>> xs = [
array('b', [99, 111, 117, 110, 116, 45, 105, 116, 101, 109, 115]),
array('b', [99, 111, 117, 110, 116, 45, 105, 116, 101, 109, 115]),
array('b', [98, 121, 116, 101, 45, 115, 105, 122, 101])
]

>>> xs[0] == xs[1]
True
>>> xs[0] == xs[2]
False
>>> map(hash, xs)
[1363822052, 1722768342, 576563150]

Just missing the appropriate support so that Object#hashCode is not invoked (the hash code is defined for all objects in Java, unlike Python). Simply following the example of PyByteArray (which implements bytearray), we get the following patch:

+    @Override
+    public int hashCode()
+        return array___hash__();
+    }
+
+    @ExposedMethod
+    final int array___hash__() {
+        throw Py.TypeError(String.format("unhashable type: '%.200s'", getType().fastGetName()));
+    }
+

Just need to add a test to complete.

Should have included the header as well - that patch is against org.python.core.PyArray

Fixed as of https://hg.python.org/jython/rev/49d498968f34

Note this means you still cannot use array.array in a dict as keys, just like CPython.

As long as it's explicit about it, that sounds fine to me,
thanks for fixing this.