Exception::Class::TryCatch - Syntactic try/catch sugar for use with Exception::Class

This documentation describes version 1.12.

     use Exception::Class::TryCatch;

     # simple usage of catch()

     eval { Exception::Class::Base->throw('error') };
     catch my $err and warn $err->error;

     # catching only certain types or else rethrowing

     eval { Exception::Class::Base::SubClass->throw('error') };
     catch( my $err, ['Exception::Class::Base', 'Other::Exception'] )
         and warn $err->error; 

     # catching and handling different types of errors

     eval { Exception::Class::Base->throw('error') };
     if ( catch my $err ) {
         $err->isa('this') and do { handle_this($err) };
         $err->isa('that') and do { handle_that($err) };
     }

     # use "try eval" to push exceptions onto a stack to catch later

     try eval { 
         Exception::Class::Base->throw('error') 
     };
     do {
         # cleanup that might use "try/catch" again
     };
     catch my $err; # catches a matching "try"

Exception::Class::TryCatch provides syntactic sugar for use with Exception::Class using the familiar keywords try and catch. Its primary objective is to allow users to avoid dealing directly with $@ by ensuring that any exceptions caught in an eval are captured as Exception::Class objects, whether they were thrown objects to begin with or whether the error resulted from die. This means that users may immediately use isa and various Exception::Class methods to process the exception.

In addition, this module provides for a method to push errors onto a hidden error stack immediately after an eval so that cleanup code or other error handling may also call eval without the original error in $@ being lost.

Inspiration for this module is due in part to Dave Rolsky's article "Exception Handling in Perl With Exception::Class" in The Perl Journal (Rolsky 2004).

The try/catch syntax used in this module does not use code reference prototypes the way the Error.pm module does, but simply provides some helpful functionality when used in combination with eval. As a result, it avoids the complexity and dangers involving nested closures and memory leaks inherent in Error.pm (Perrin 2003).

Rolsky (2004) notes that these memory leaks may not occur in recent versions of Perl, but the approach used in Exception::Class::TryCatch should be safe for all versions of Perl as it leaves all code execution to the eval in the current scope, avoiding closures altogether.

     # zero argument form
     my $err = catch;

     # one argument forms
     catch my $err;
     my $err = catch( [ 'Exception::Type', 'Exception::Other::Type' ] );

     # two argument form
     catch my $err, [ 'Exception::Type', 'Exception::Other::Type' ];

Returns an Exception::Class::Base object (or an object which is a subclass of it) if an exception has been caught by eval. If no exception was thrown, it returns undef in scalar context and an empty list in list context. The exception is either popped from a hidden error stack (see try) or, if the stack is empty, taken from the current value of $@.

If the exception is not an Exception::Class::Base object (or subclass object), an Exception::Class::Base object will be created using the string contents of the exception. This means that calls to die will be wrapped and may be treated as exception objects. Other objects caught will be stringfied and wrapped likewise. Such wrapping will likely result in confusing stack traces and the like, so any methods other than error used on Exception::Class::Base objects caught should be used with caution.

catch is prototyped to take up to two optional scalar arguments. The single argument form has two variations.

• If the argument is a reference to an array, any exception caught that is not of the same type (or a subtype) of one of the classes listed in the array will be rethrown.
• If the argument is not a reference to an array, catch will set the argument to the same value that is returned. This allows for the catch my $err idiom without parentheses.

In the two-argument form, the first argument is set to the same value as is returned. The second argument must be an array reference and is handled the same as as for the single argument version with an array reference, as given above.

caught is a synonym for catch for syntactic convenience.

NOTE: Exception::Class version 1.21 added a "caught" method of its own. It provides somewhat similar functionality to this subroutine, but with very different semantics. As this class is intended to work closely with Exception::Class, the existence of a subroutine and a method with the same name is liable to cause confusion and this method is deprecated and may be removed in future releases of Exception::Class::TryCatch.

This method is no longer exported by default.

     # void context
     try eval {
       # dangerous code
     };
     do {
       # cleanup code can use try/catch
     };
     catch my $err;

     # scalar context
     $rv = try eval { return $scalar };

     # list context
     @rv = try [ eval { return @array } ];

Pushes the current error ($@) onto a hidden error stack for later use by catch. try uses a prototype that expects a single scalar so that it can be used with eval without parentheses. As eval { BLOCK } is an argument to try, it will be evaluated just prior to try, ensuring that try captures the correct error status. try does not itself handle any errors -- it merely records the results of eval. try { BLOCK } will be interpreted as passing a hash reference and will (probably) not compile. (And if it does, it will result in very unexpected behavior.)

Since try requires a single argument, eval will normally be called in scalar context. To use eval in list context with try, put the call to eval in an anonymous array:

   @rv = try [ eval {return @array} ];

When try is called in list context, if the argument to try is an array reference, try will dereference the array and return the resulting list.

In scalar context, try passes through the scalar value returned by eval without modifications -- even if that is an array reference.

   $rv = try eval { return $scalar };
   $rv = try eval { return [ qw( anonymous array ) ] };

Of course, if the eval throws an exception, eval and thus try will return undef.

try must always be properly bracketed with a matching catch or unexpected behavior may result when catch pops the error off of the stack. try executes right after its eval, so inconsistent usage of try like the following will work as expected:

     try eval {
         eval { die "inner" };
         catch my $inner_err
         die "outer" if $inner_err;
     };
     catch my $outer_err;
     # handle $outer_err;

However, the following code is a problem:

     # BAD EXAMPLE
     try eval {
         try eval { die "inner" };
         die $@ if $@;
     };
     catch my $outer_err;
     # handle $outer_err;

This code will appear to run correctly, but catch gets the exception from the inner try, not the outer one, and there will still be an exception on the error stack which will be caught by the next catch in the program, causing unexpected (and likely hard to track) behavior.

In short, if you use try, you must have a matching catch. The problem code above should be rewritten as:

     try eval {
         try eval { die "inner" };
         catch my $inner_err;
         $inner_err->rethrow if $inner_err;
     };
     catch my $outer_err;
     # handle $outer_err;

Please report any bugs or feature using the CPAN Request Tracker. Bugs can be submitted through the web interface at http://rt.cpan.org/Dist/Display.html?Queue=Exception-Class-TryCatch

When submitting a bug or request, please include a test-file or a patch to an existing test-file that illustrates the bug or desired feature.

1.

perrin. (2003), "Re: Re2: Learning how to use the Error module by example", (perlmonks.org), Available: http://www.perlmonks.org/index.pl?node_id=278900 (Accessed September 8, 2004).

2.

Rolsky, D. (2004), "Exception Handling in Perl with Exception::Class", The Perl Journal, vol. 8, no. 7, pp. 9-13

• Exception::Class
• Error -- but see (Perrin 2003) before using

David A. Golden (DAGOLDEN)

Copyright (c) 2004-2008 by David A. Golden. All rights reserved.

Licensed under Apache License, Version 2.0 (the "License"). You may not use this file except in compliance with the License. A copy of the License was distributed with this file or you may obtain a copy of the License from http://www.apache.org/licenses/LICENSE-2.0

Files produced as output though the use of this software, shall not be considered Derivative Works, but shall be considered the original work of the Licensor.

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.