Structured Exception Handling in Visual Studio .NET
Structured Exception Handling in Visual Studio.Net
By Peter Aitken
June 17, 2004
For programmers, exceptions—that's just a fancy name for run-time errors—are a fact of life. Good programming practices can minimize exceptions, but it is impossible to prevent them altogether. Many of the causes, such as network errors or incorrectly set file permissions, are beyond the programmer's control.
You can, however, handle exceptions in your program. No, let me rephrase that—you must handle exceptions in your program! Fortunately, the .Net framework and languages include powerful exception handling tools. I'll use Visual Basic, in this article, but the same principles and techniques apply to C# as well.
An exception is a problem that occurs up while the program is executing. In .Net and object-oriented terminology, a program is said to throw an exception. Many situations can cause exceptions, including trying to open a non-existent file for reading, attempting to instantiate a class that is not available, or writing to a removable media drive when no media is inserted. If an exception is thrown that the program does not handle, a message is displayed to the user, and the program usually ends (although with some exceptions, continuing is also an option). This is a polite way of saying that the program crashes, and believe me: that's no way for a programmer to impress clients!
The .Net framework implements structured exception handling. The term structured means that the code that has the potential to cause an exception. Plus, the code to handle the exception is organized in structured blocks. You use the Try...Catch...End Try
statement for this purpose. The syntax is as follows:
Try ' Executable statements that may cause an exception are placed here. Catch [optional filters] ' Statements that are to be executed when an exception occurs go here. ' The optional filters permit the code to respond only to specific ' types of exceptions. You can place multiple Catch blocks in a ' Try statement. Finally ' Code placed here is executed after the code in the Try block ' or in the Catch block is executed. This block is optional. End Try
Structured exception handling operates on the level of procedures. Each procedure in your program should contain exception handling code designed to catch the types of exceptions that the procedure might generate. If a procedure generates an unhandled exception, it is passed up the call stack to the caller. In other words, if procedure A calls procedure B, and procedure B generates an unhandled exception, it is passed up to procedure A for handling.
The ability to filter the Catch
statement is an essential part of .Net's structured exception handling. If there is no filter—or, to be more precise, there is a filter that matches all exceptions—the code in the Catch
block is executed for any and all exceptions that are thrown. Such a Catch
block would be written like this:
Catch ex As Exception
Within the Catch
block, the object ex
, an instance of the System.Exception
class, will contain information about the exception. You can use any legal variable name here, although ex
or e
is traditional. To filter for specific exceptions, you would write the Catch
statement like this:
Catch ex As ExceptionType
replacing ExceptionType
with the name of the specific exception class to be caught. The .Net framework defines a hierarchy of several dozen exception classes, all derived from System.Exception
, with each individual exception class encompassing related exceptions. For example, all file-related exceptions are subsumed under IoException
, but for more specific control you can catch file loading exceptions with FileLoadException
and FileNotFoundException
. You'll find a complete diagram of the exception hierarchy in the Visual Studio documentation.
The filtering capability lets you write Catch
blocks to deal with specific exceptions. This makes perfect sense, because the message to the user and the program's response will differ, depending on the type of exception. Simply create a separate Catch
block for each condition. It is also common practice to include a final Catch
block to deal with any exceptions that fall through the preceding Catch filters. Here's an example:
Catch ex As FileNotFoundException ' Code here to respond to a "File Not Found" exception. Catch ex As FileLoadException ' Code here to respond to a "File Load" exception. Catch ex As Exception ' Code here to respond to any exceptions not caught by previous Catch statements.
The Catch
keyword, by itself, will catch any exception, just like Catch ex As Exception
. The disadvantage is that you will not have an Exception
object in that block to provide information about the exception and its cause (as will be explained below).
In addition to filtering exceptions based on the exception type, you can perform additional filtering using the When
clause in the Catch
statement. The syntax is:
Catch ex As ExceptionType When Expression
Expression
is any expression that evaluates to True or False. The Catch
block is executed only if the exception that was thrown matches ExceptionType and Expression is True. This lets you execute one of two or more Catch
blocks for the same exception, depending on program conditions. Here's an example:
Try ' Code that might cause an OverflowException Catch ex As OverflowException When Newuser = True ' Display error info for new users. Catch ex As OverflowException When Newuser = False ' Display error info for experienced users.
What can your program do to respond to an exception? Of course, the details depend on your program and the specific exception, but you need some information to work with. All exception classes have a base set of properties that provide some information about the exception, as follows:
Property | Description |
HelpLink | Specifies the Uniform Resource Name (URN) or Uniform Resource Locater (URL) of the help file associated with this exception. |
Message | A short description of the exception. |
Source | The name of the application or object that caused the exception |
StackTrace | Identifies the stack trace–the sequence of procedure calls leading up to the location in code where the exception occurred. |
Some derived exception classes have additional properties specific to the associated
exception. For example, the FileLoadException
class, which is thrown
when there is an error loading a file, has the FileName
property;
it provides the name of the file that you were attempting to load. You'll find
details on these specific exception members in the .Net framework documentation.
What about the optional Finally
block? It is always executed, no matter what happens in the Try
and Catch
sections. Use it for actions that have to be performed regardless of whether an exception was thrown, such as closing files. The Finally
block is executed even if the thrown exception was not caught by any of the Catch
statements.
The .Net framework also lets you throw your own exceptions. Why would you want to do this? You might want to test your exception handling code before compiling and distributing the program. Another reason is to create and use custom exceptions, so your program can make use of VB.Net's exception handling mechanism to handle program-specific error conditions (such as the user entering incorrect data).
Also, you may want to throw an exception inside a Catch
block, when your code's response to a caught exception is to throw a new exception or to re-throw the original exception. This is sometimes desirable when handling exceptions in a component. You may want the code in the component to take some action in response to the exception while, at the same time, passing the exception back up the call stack for additional handling.
To throw an exception, use the Throw
statement. The syntax is:
Throw ex
where ex is an object of type Exception
. For example:
Throw New FileNotFoundException ("The file you requested does not exist.")
If used within a Catch
block, you can use Throw
with no argument to re-throw the exception that was caught originally.
Finally, Visual Basic.Net still supports unstructured exception handling. This is the older style of error handling that was available in Visual Basic version 6 and earlier, using the On Error Goto, Resume
, and related statements. It is supported primarily for compatibility with legacy Visual Basic code; I see no reason to use it in new code. You can use both structured and unstructured exception handling in a program, but each procedure is limited to one or the other.
Exception handling is an integral part of every well-written program. There's no excuse for letting unhandled exceptions bedevil your end-users. I recommend that you built it in from the start. It's a lot easier than trying to go back to add it later.