Anonymous and local class names changed in 1.5

[Mike Schilling]

Consider the Java class::

class Outer {
class Inner {
Object o = new Object() {
public String toString() { return "MYOB";}};
void method() {
class Local {}
}
}
}

In 1.4.2, there are classes Outer$1.class and Outer$1Local.class. In 1.5,
these become Outer$Inner$1.class and Outer$Inner$1Local.class. Can anyone
see a reason for this other than to annoy me (and break the tools I use to
analyze Java classes)?

 

[zero]

Consider the Java class::

class Outer {
class Inner {
Object o = new Object() {
public String toString() { return "MYOB";}};
void method() {
class Local {}
}
}
}

In 1.4.2, there are classes Outer$1.class and Outer$1Local.class. In
1.5, these become Outer$Inner$1.class and Outer$Inner$1Local.class.
Can anyone see a reason for this other than to annoy me (and break the
tools I use to analyze Java classes)?

The 1.5 way seems more logical and natural to me. After all your anonymous
and Local classes are inside Inner inside Outer. Maybe the guys at Sun
thought so too.

 

[Thomas Hawtin]

In 1.4.2, there are classes Outer$1.class and Outer$1Local.class. In 1.5,
these become Outer$Inner$1.class and Outer$Inner$1Local.class. Can anyone
see a reason for this other than to annoy me (and break the tools I use to
analyze Java classes)?

The good news is the JLS3 gives rules to what the format should be, so
shouldn't change too much in the future.

Tom Hawtin

 

[Mike Schilling]

The good news is the JLS3 gives rules to what the format should be, so
shouldn't change too much in the future.

So it does.
(http://java.sun.com/docs/books/jls/third_edition/html/binaryComp.html#13.1)
Thanks!

 

[Roedy Green]

In 1.4.2, there are classes Outer$1.class and Outer$1Local.class. In 1.5,
these become Outer$Inner$1.class and Outer$Inner$1Local.class. Can anyone
see a reason for this other than to annoy me (and break the tools I use to
analyze Java classes)?

A bug has been corrected.

 

[Mike Schilling]

A bug has been corrected.

I don't see any bug. The change to local class names in 1.4.2 *was* a
bugfix, IIRC, because the previous names could be ambiguous, but the 1.4.2.
naming convention worked perfectly well.

 

[Roedy Green]

I don't see any bug. The change to local class names in 1.4.2 *was* a
bugfix, IIRC, because the previous names could be ambiguous, but the 1.4.2.
naming convention worked perfectly well.

To me they seem wrong if that description of the names is correct. A
name should reflect something's proper place in the hierarchy. (how
feudal!) In any case the designers were putting the naming rules in
the JLS so they probably decided to tidy them up to make them easier
to explain and easier to defend.