有如下代码:
procedure Change(a:array of Int32);
procedure Change0(var a:array of Int32);
type
Ta = array[0..10] of Int32;
procedure Change2(a:Ta);
procedure Change3(var a:Ta);
var
aa:Ta;
bb:array of Int32;
implementation
{$AUTOBOX ON}
procedure Change(a:array of Int32);
begin
a[0]:=123;
a:=bb;
end;
procedure Change0(var a:array of Int32);
begin
a[0]:=123;
a:=bb;
end;
procedure Change2(a:Ta);
begin
a[0]:=123;
a:=bb;
end;
procedure Change3(var a:Ta);
begin
a[0]:=123;
a:=bb;
end;
然后执行下面的语句
SetLength(bb,11);bb[0]:=5678;
aa := Ta(&Array.CreateInstance(typeof(Int32),11));
aa[0]:=0;
Change(aa);
MessageBox.Show(aa[0].ToString());
if &Object(aa)=&Object(bb) then MessageBox.Show('=');
aa := Ta(&Array.CreateInstance(typeof(Int32),11));
aa[0]:=0;
Change0(aa);
MessageBox.Show(aa[0].ToString());
if &Object(aa)=&Object(bb) then MessageBox.Show('=');
aa := Ta(&Array.CreateInstance(typeof(Int32),11));
aa[0]:=0;
Change2(aa);
MessageBox.Show(aa[0].ToString());
if &Object(aa)=&Object(bb) then MessageBox.Show('=');
aa := Ta(&Array.CreateInstance(typeof(Int32),11));
aa[0]:=0;
Change3(aa);
MessageBox.Show(aa[0].ToString());
if &Object(aa)=&Object(bb) then MessageBox.Show('=');
结果发现 array of Int32 方式,可以改变数组元素的值,但不能改变数组变量中存储的数组首地址,输出123
var array of Int32 既可以改变数组的值,又可以改变数组变量中存储的数组首地址,输出5678和=
Ta方式,不能改变数组元素的值,但是却很奇怪,aa和bb指向同一个数组,输出0和=
var Ta方式 可以改变数组元素的值,但是却不是5678而是123;但是aa和bb指向同一个数组,输出123和=
反汇编结果如下
method public static void Change(int32[] a) cil managed
{
// Code Size: 13 byte(s)
.maxstack 3
L_0000: ldarg.0
L_0001: ldc.i4.0
L_0002: ldc.i4.s 123
L_0004: stelem.i4
L_0005: ldsfld int32[] WinForm.Units.WinForm::bb
L_000a: starg.s a
L_000c: ret
}
.method public static void Change0(int32[]& a) cil managed { // Code Size: 14 byte(s) .maxstack 3 L_0000: ldarg.0 L_0001: ldind.ref L_0002: ldc.i4.0 L_0003: ldc.i4.s 123 L_0005: stelem.i4 L_0006: ldarg.0 L_0007: ldsfld int32[] WinForm.Units.WinForm::bb L_000c: stind.ref L_000d: ret }
.method public static void Change2(int32[] a) cil managed
{
// Code Size: 42 byte(s)
.maxstack 4
.locals (
int32 num1)
L_0000: ldarg.0
L_0001: callvirt instance object [mscorlib]System.Array::Clone()
L_0006: castclass int32[]
L_000b: starg.s a
L_000d: ldarg.0
L_000e: ldc.i4.0
L_000f: ldc.i4.s 123
L_0011: stelem.i4
L_0012: ldsfld int32[] WinForm.Units.WinForm::bb
L_0017: dup
L_0018: ldlen
L_0019: stloc.0
L_001a: ldarg.0
L_001b: ldloc.0
L_001c: ldc.i4.s 11
L_001e: ble.s L_0023
L_0020: ldc.i4.s 11
L_0022: stloc.0
L_0023: ldloc.0
L_0024: call void [mscorlib]System.Array::Copy([mscorlib]System.Array, [mscorlib]System.Array, int32)
L_0029: ret
}
.method public static void Change3(int32[]& a) cil managed
{
// Code Size: 31 byte(s)
.maxstack 4
.locals (
int32 num1)
L_0000: ldarg.0
L_0001: ldind.ref
L_0002: ldc.i4.0
L_0003: ldc.i4.s 123
L_0005: stelem.i4
L_0006: ldsfld int32[] WinForm.Units.WinForm::bb
L_000b: dup
L_000c: ldlen
L_000d: stloc.0
L_000e: ldarg.0
L_000f: ldind.ref
L_0010: ldloc.0
L_0011: ldc.i4.s 11
L_0013: ble.s L_0018
L_0015: ldc.i4.s 11
L_0017: stloc.0
L_0018: ldloc.0
L_0019: call void [mscorlib]System.Array::Copy([mscorlib]System.Array, [mscorlib]System.Array, int32)
L_001e: ret
}
结论:
使用array of Int32 方式,实际上对应于C#的可变数目参数,即 params Int32[],
这种参数方式是传递的数组的首地址(即参数的值),而不是存放数组首地址的变量的地址
加上Var修饰,即加入ref修饰,传递的是存放数组首地址的变量的地址(即参数变量自身的地址)
使用Ta方式,则在函数内部对数组的值进行克隆,既不改变数组的首地址,也不改变原数组的值
加入var 修饰,好像传递的是变量自身的地址,这里还是没看懂是为什么。
对应的C#代码为
public static void Change(params int[] a)
{
a[0] = 0x7b;
a = WinForm.bb;
}
public static void Change0(params ref int[] a)
{
a[0] = 0x7b;
a = WinForm.bb;
}
public static void Change2(int[] a)
{
a = (int[]) a.Clone();
a[0] = 0x7b;
int num1 = WinForm.bb.Length;
if (num1 > 11)
{
num1 = 11;
}
Array.Copy(WinForm.bb, a, num1);
}
public static void Change3(ref int[] a)
{
a[0] = 0x7b;
int num1 = WinForm.bb.Length;
if (num1 > 11)
{
num1 = 11;
}
Array.Copy(WinForm.bb, a, num1);
}
本文地址:http://com.8s8s.com/it/it4541.htm