无法访问的代码，但可以通过异常访问

此代码是从 ODBC 连接的数据库读取和写入的应用程序的一部分。它在数据库中创建一条记录，然后检查是否成功创建了一条记录，然后返回 true。

我对控制流的理解如下:

当“一个方法调用对于对象的当前状态无效”时，command.ExecuteNonQuery()会抛出一个 Invalid​Operation​Exception。因此，如果发生这种情况，try块的执行将停止，finally块将被执行，然后在底部执行 return false;。

但是，我的 IDE 声称 return false;是无法访问的代码。似乎是真的，我可以删除它，它编译没有任何抱怨。但是，在我看来，抛出上述异常的代码路径似乎没有返回值。

private static bool createRecord(String table,
IDictionary<String,String> data,
System.Data.IDbConnection conn,
OdbcTransaction trans) {


[... some other code ...]


int returnValue = 0;
try {
command.CommandText = sb.ToString();
returnValue = command.ExecuteNonQuery();


return returnValue == 1;
} finally {
command.Dispose();
}


return false;
}

我的理解有什么错误吗？

16745 次浏览
小开

the finally block would be executed, then would execute the return false; at the bottom.

Wrong. finally doesn't swallow the exception. It honors it and the exception will be thrown as normal. It will only execute the code in the finally before the block ends (with or without an exception).

If you want the exception to be swallowed, you should use a catch block with no throw in it.

小开

When the exception is thrown, the stack will unwind (execution will move out of the function) without returning a value, and any catch block in the stack frames above the function will catch the exception instead.

Hence, return false will never execute.

Try manually throwing an exception to understand the control flow:

try {
command.CommandText = sb.ToString();
returnValue = command.ExecuteNonQuery();


// Try this.
throw new Exception("See where this goes.");


return returnValue == 1;
} finally {
command.Dispose();
}
小开
最佳答案

Compiler Warning (level 2) CS0162

Unreachable code detected

The compiler detected code that will never be executed.

Which is just saying, the Compiler understands enough through Static Analysis that it can't be reached and completely omits it from the compiled IL (hence your warning).

Note : You can prove this fact to your self by trying to Step on to the Unreachable Code with the debugger, or using an IL Explorer.

The finally may run on an Exception, (though that aside) it doesn't change the fact (in this case) it will still be an Uncaught Exception. Ergo, the last return will never get hit regardless.

  • If you want the code to continue onto the last return, your only option is to Catch the Exception;

  • If you don't, just leave it the way it is and remove the return.

Example

try
{
command.CommandText = sb.ToString();
returnValue = command.ExecuteNonQuery();


return returnValue == 1;
}
catch(<some exception>)
{
// do something
}
finally
{
command.Dispose();
}


return false;

To quote the documentation

try-finally (C# Reference)

By using a finally block, you can clean up any resources that are allocated in a try block, and you can run code even if an exception occurs in the try block. Typically, the statements of a finally block run when control leaves a try statement. The transfer of control can occur as a result of normal execution, of execution of a break, continue, goto, or return statement, or of propagation of an exception out of the try statement.

Within a handled exception, the associated finally block is guaranteed to be run. However, if the exception is unhandled, execution of the finally block is dependent on how the exception unwind operation is triggered. That, in turn, is dependent on how your computer is set up.

Usually, when an unhandled exception ends an application, whether or not the finally block is run is not important. However, if you have statements in a finally block that must be run even in that situation, one solution is to add a catch block to the try-finally statement. Alternatively, you can catch the exception that might be thrown in the try block of a try-finally statement higher up the call stack. That is, you can catch the exception in the method that calls the method that contains the try-finally statement, or in the method that calls that method, or in any method in the call stack. If the exception is not caught, execution of the finally block depends on whether the operating system chooses to trigger an exception unwind operation.

Lastly

When using anything that supports the IDisposable interface (which is designed to release unmanaged resources), you can wrap it in a using statement. The compiler will generate a try {} finally {} and internally call Dispose() on the object.

小开

The last statement return false is unreachable, because the try block is missing a catch part that would handle the exception, so the exception is rethrown after the finally block and the execution never reaches the last statement.

小开

On your code:

private static bool createRecord(String table, IDictionary<String,String> data, System.Data.IDbConnection conn, OdbcTransaction trans) {


[... some other code ...]


int returnValue = 0;
try {
command.CommandText = sb.ToString();
returnValue = command.ExecuteNonQuery();


return returnValue == 1; // You return here in case no exception is thrown
} finally {
command.Dispose(); //You don't have a catch so the exception is passed on if thrown
}


return false; // This is never executed because there was either one of the above two exit points of the method reached.
}

the finally block would be executed, then would execute the return false; at the bottom

This is the flaw in your logic because the finally block won't catch the exception and it will never reach the last return statement.

小开

It seems, you are looking for something like this:

private static bool createRecord(string table,
IDictionary<String,String> data,
System.Data.IDbConnection conn,
OdbcTransaction trans) {
[... some other code ...]


// Using: do not call Dispose() explicitly, but wrap IDisposable into using
using (var command = ...) {
try {
// Normal flow:
command.CommandText = sb.ToString();


// True if and only if exactly one record affected
return command.ExecuteNonQuery() == 1;
}
catch (DbException) {
// Exceptional flow (all database exceptions)
return false;
}
}
}

Please, note, that finally doesn't swallow any exception

finally {
// This code will be executed; the exception will be efficently re-thrown
}


// And this code will never be reached
小开

The warning is because you didn't use catch and your method is basically written like this:

bool SomeMethod()
{
return true;
return false; // CS0162 Unreachable code detected
}

Since you use finally solely to dispose, the preferred solution is to utilize using pattern:

using(var command = new WhateverCommand())
{
...
}

That's enough, to ensure what Dispose will be called. It's guaranteed to be called either after successful execution of code block or upon (before) some catch down in call stack (parent calls are down, right?).

If it wouldn't be about disposing, then 

try { ...; return true; } // only one return
finally { ... }

is enough, since you will never have to return false at the end of method (there is no need for that line). Your method is either return result of command execution (true or false) or will throw an exception otherwise.

Consider also to throw own exceptions by wrapping expected exceptions (check out InvalidOperationException constructor):

try { ... }
catch(SomeExpectedException e)
{
throw new SomeBetterExceptionWithExplanaition("...", e);
}

This is typically used to say something more meaningful (useful) to the caller than nested call exception would be telling.

Most of times you don't really care about unhandled exceptions. Sometimes you need to ensure that finally is called even if exception is unhandled. In this case you simply catch it yourself and re-throw (see this answer):

try { ... }
catch { ...; throw; } // re-throw
finally { ... }
小开

You don't have a catch block, so the exception is still thrown, which blocks the return.

the finally block would be executed, then would execute the return false; at the bottom.

This is wrong, because the finally block would be executed, and then there would be an uncaught exception.

finally blocks are used for cleanup, and they do not catch the exception. The exception is thrown before the return, therefore, the return will never be reached, because an exception is thrown before.

Your IDE is correct that it will never be reached, because the exception will be thrown. Only catch blocks are able to catch exceptions.

Reading from the documentation,

Usually, when an unhandled exception ends an application, whether or not the finally block is run is not important. However, if you have statements in a finally block that must be run even in that situation, one solution is to add a catch block to the try-finally statement. Alternatively, you can catch the exception that might be thrown in the try block of a try-finally statement higher up the call stack. That is, you can catch the exception in the method that calls the method that contains the try-finally statement, or in the method that calls that method, or in any method in the call stack. If the exception is not caught, execution of the finally block depends on whether the operating system chooses to trigger an exception unwind operation.

This clearly shows that the finally is not intended to catch the exception, and you would have been correct if there had been an empty catch statement before the finally statement.

小开

You have two return paths in your code, the second of which is unreachable because of the first. The last statement in your try block return returnValue == 1; provides your normal return, so you can never reach the return false; at the end of the method block.

FWIW, order of exection related to the finally block is: the expression supplying the return value in the try block will be evaluated first, then the finally block will be executed, and then the calculated expression value will be returned (inside the try block).

Regarding flow on exception... without a catch, the finally will be executed upon exception before the exception is then rethrown out of the method; there is no "return" path.


提交答案