bug
bug copied to clipboard
Mishandled captured parameters in secondary class constructors.
There are some conditions where parameter captured by secondary constructors generate ambiguous bytecode. Such as in the following example where the default constructor receives an Int parameter and the secondary captures an Int.
object Bar extends App {
val
def foo(x: Int) = {
class Foo(i: Int) {
def this() = this(x)
}
new Foo
}
println(foo(6))
}
Will generate the following bytecode where the two constructors have the same signature.
public class Bar$Foo$1 {
public Bar$Foo$1(int);
descriptor: (I)V
Code:
0: aload_0
1: invokespecial #15 // Method java/lang/Object."<init>":()V
4: return
public Bar$Foo$1(int);
descriptor: (I)V
Code:
0: aload_0
1: iload_1
2: invokespecial #21 // Method "<init>":(I)V
5: return
}
The problem does not show itself at compilation time, it only fails at runtime with:
[info] Running Bar
[error] (run-main-b) java.lang.ClassFormatError: Duplicate method name&signature in class file Bar$Foo$1
java.lang.ClassFormatError: Duplicate method name&signature in class file Bar$Foo$1
at java.lang.ClassLoader.defineClass1(Native Method)
at java.lang.ClassLoader.defineClass(ClassLoader.java:760)
at java.security.SecureClassLoader.defineClass(SecureClassLoader.java:142)
at java.net.URLClassLoader.defineClass(URLClassLoader.java:467)
at java.net.URLClassLoader.access$100(URLClassLoader.java:73)
at java.net.URLClassLoader$1.run(URLClassLoader.java:368)
at java.net.URLClassLoader$1.run(URLClassLoader.java:362)
at java.security.AccessController.doPrivileged(Native Method)
at java.net.URLClassLoader.findClass(URLClassLoader.java:361)
at java.lang.ClassLoader.loadClass(ClassLoader.java:424)
at java.lang.ClassLoader.loadClass(ClassLoader.java:357)
at Bar$.foo(A.scala:7)
at Bar$.delayedEndpoint$App$1(Bar.scala:10)
at Bar$delayedInit$body.apply(Bar.scala:1)
at scala.Function0$class.apply$mcV$sp(Function0.scala:34)
at scala.runtime.AbstractFunction0.apply$mcV$sp(AbstractFunction0.scala:12)
at scala.Bar$$anonfun$main$1.apply(App.scala:76)
at scala.Bar$$anonfun$main$1.apply(App.scala:76)
at scala.collection.immutable.List.foreach(List.scala:381)
at scala.collection.generic.TraversableForwarder$class.foreach(TraversableForwarder.scala:35)
at scala.App$class.main(App.scala:76)
at Bar$.main(Bar.scala:1)
at Bar.main(Bar.scala)
at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
at java.lang.reflect.Method.invoke(Method.java:497)
[trace] Stack trace suppressed: run last compile:run for the full output.
java.lang.RuntimeException: Nonzero exit code: 1
at scala.sys.package$.error(package.scala:27)
[trace] Stack trace suppressed: run last compile:run for the full output.
[error] (compile:run) Nonzero exit code: 1
[error] Total time: 1 s, completed Oct 26, 2015 4:42:21 PM
The compiler should handle these cases in a different way or should disallow them by failing the compilation.
Imported From: https://issues.scala-lang.org/browse/SI-9536?orig=1 Reporter: Nicolas Stucki (nicolas.stucki) Affected Versions: 2.10.5, 2.11.7, 2.12.0-M3
@retronym said: Great test case!
Lambdalift basically assumes a world without overloading, a restriction which is enforced for local methods:
def foo { def bar(a: Int) = 0; def bar(a: String) = 0 } // error: method bar is defined twice
However, this is not enforced for constructors of local classes, which are handled by the same code in lambdalift (see addFreeParams).
One possible resolution of this bug would be to enforce such a restriction (although it would need to be phased with a deprecation warning)
An analagous condundrum exists in the Java compiler when it needs to add constructors to a class to allow inner classes to invoke a private enclosing constructor. Java adds dummy parameter to the method to avoid ambiguity with existing constructors. See https://github.com/scala/scala/pull/2750 for some more details. This approach might also work here, but I don't think the cost/benefit stacks up.
The last reference on Jan 16 was a typo, but editing the comment did not cause this "notification of reference" to disappear.
Not yet fixed on Scala 3. It probably would have been fixed by now if they hadn't removed the quickfix label.
Fixed in 3.3.1. The nested class gets an outer pointer, where the sigs are (outer, int) and (int, outer). I couldn't find the change with just a dart and a blindfold.