| Re: [jdt-dev] How does this code work ? |
|
While that mystery is solved, the remaining puzzle is how we ended up with two nearly identical versions of the method
generateClassLiteralAccessForType in CodeStream and StackMapFrameCodeStream.
Srikanth From:
jdt-dev <jdt-dev-bounces@xxxxxxxxxxx> On Behalf Of Sankaran, Srikanth via jdt-dev CAUTION:
This email was sent from outside of Advantest. It seems you need a branch and so a different stackmap frame for verification to catch such problems. I have raised
https://github.com/eclipse-jdt/eclipse.jdt.core/issues/2476 with a test case where we do get a verify error. Thanks Srikanth From:
jdt-dev <jdt-dev-bounces@xxxxxxxxxxx>
On Behalf Of Sankaran, Srikanth via jdt-dev CAUTION:
This email was sent from outside of Advantest. I have been staring at this code for a while now wondering how the bad code generated still works: public
class X {
public
static
void main(String[]
args) {
foo(String.class,
switch (42) {
default -> {
try {
yield 42; }
finally { } } }); }
static
void foo(Class<?>
c,
int
v) { System.out.println(c); System.out.println(v); } } On master, we generate the following code: public static void main(java.lang.String[]); descriptor: ([Ljava/lang/String;)V flags: (0x0009) ACC_PUBLIC, ACC_STATIC Code: stack=2, locals=3, args_size=1 0: ldc #16 // class java/lang/String 2: astore_1 <<ß-------------------- Class Object being stored into String See LVT slot 1 below. 3: bipush 42 5: istore_2 6: aload_1 <<ß-------------------- aload loads an object statically typed String, but at bci 8 we invoke a method that expects a Class object 7: iload_2 8: invokestatic #18 // Method foo:(Ljava/lang/Class;I)V 11: return LineNumberTable: line 4: 0 line 7: 3 line 4: 8 line 13: 11 LocalVariableTable: Start Length Slot Name Signature 0 12 0 args [Ljava/lang/String; 3 5 1 yieldValue0 Ljava/lang/String; <<ß---------------------- String typed local 6 2 2 secretYieldValue I What am I overlooking ?? Srikanth |