Name: gm110360 Date: 03/19/2002


FULL PRODUCT VERSION :
All versions

A DESCRIPTION OF THE PROBLEM :
I suggest canceling the hidden narrowing casting in the
compound assignments such as +=.

Currently the JLS states:
"A compound assignment expression of the form E1 op = E2 is
equivalent to E1 = (T )((E1) op (E2)), where T is the type
of E1, except that E1 is evaluated only once. Note that the
implied cast to type T may be either an identity conversion
(§5.1.1) or a narrowing primitive conversion (§5.1.3)."

Narrowing casting being an unsafe thing by itself and
requiring a thought each time it?s used, is implicit and
hidden here, which makes it much more dangerous.

There is no other place (IMHO) in Java where compiler
performs a data-wiping casting without the programmer's
knowledge and an explicit approval. Thus this feature is
inconsistent with the strong typing system generally
exhibited by Java and compromises its type safety.

Moreover, most, if not all, Java programming and tutorial
books state that a+=b is equivalent to a=a+b. So if one day
they will become equal nobody will pay attention except
those who will be surprised to find bugs hidden in their
code due to the feature. The negligible minority who used
the feature consciously will have to rewrite it as a=(T)
(a+b), which is a small price for regaining Java type
safety.


STEPS TO FOLLOW TO REPRODUCE THE PROBLEM :
int x = 0;
x += 0.5;

EXPECTED VERSUS ACTUAL BEHAVIOR :
I would expect an error here or at least a warning.

This bug can be reproduced always.
(Review ID: 139922)
======================================================================