NAME errors - Error Handling for Perl STATUS This module is still under design. Don't use it in production yet. SYNOPSIS use strict; use warnings; use errors; try { $cool = something(); throw UncoolError("Something is not cool") if not $cool; assert($ok, "Everything is ok"); } catch AssertionError with { my $e = shift; warn "$e"; } catch UncoolError with { # $@ is the same as $_[0] warn "$@"; } except { warn "Some other error: $@"; } finally { cleanup(); }; DESCRIPTION The "errors" module adds clean, simple, sane, flexible and usable error handling to Perl. The module does several things: First, "errors" exports a error handling syntax that is backwards compatible with Error.pm, but with a few improvements. Error.pm syntax is very well done; about as close to other modern language's exception handling as you can get using Pure Normal Perl. Second, all errors that are thrown are first class Perl objects. They all inherit from the "Error" class, which is provided by default. This allows you to manipulate errors in a consistent and intuitive way. Third, The "errors" module makes it trivial to define your own error classes, and encourages you to do so. Simply define a class that inherits from "Error" (or one of its subclasses). Fourth, "errors" turns plain (string based) system errors and other die/croak errors into specific Perl objects. It uses heuristics on the error string to determine which Error class to use, and defaults to the "RuntimeError" class. Fifth, "errors" provides dozens of predefined error classes that you can use or that get used automatically by the auto-objectification. These classes are in an inheritance hierarchy that should become standard for Perl. Lastly, "errors" is designed to play nice with all the modern Perl frameworks (like Moose) and the other popular error handling modules. SIMPLE TO USE The main goal of "errors" is to encourage the widespread use of error handling in Perl. In other languages like Python, coining your own named error classes and using raise/except is as common as using if/else statements. Here's a Python example. class FooError(Exception): pass try: something() catch FooError as e: handle_error(e) Now you can do that in Perl: use errors; package FooError; use base 'Error'; package MyModule; try { something(); } catch FooError with { my $e = shift; handle_error($e); }; As you can see, using "errors" is simple and unobtrusive. Why not start all your programs with: use strict; use errors; use warnings; Defining your own error classes is also trivial, and "errors" provides an even more concise way to do it: use errors -class => 'FooError'; In the catch/with clause, you can also use $@ to access the current error object like this: catch FooError with { handle_error($@); }; USAGE There are a few different usages of "errors" that you should be aware of: use errors; This exports the "errors" syntax, and loads all the "errors" functionality. use errors -class => 'ClassName' [, -isa => 'BaseClass']; The "-class" directive gives you a way to define an error subclass at compile time, in one simple line of code. You can optionally specify the base class. The default base class is "Error". NOTE: This usage does not export the "errors" (try/catch) syntax. use errors -with_using; Unfortunately "Moose" and "errors" both export the "with" subroutine. If "errors" sees that "Moose" (or someone else) has already exported "with", it will export the "using" subroutine instead: use Moose; use errors; try {...} catch Error using {...}; The "-with_using" directive tells "errors" to do this regardless. SYNTAX The "errors" module introduces a number of keyword constructs that you can use to create and manage error objects. try { ... } Like an eval block. After the code is evaluated, the appropriate clauses (catch, except, otherwise, finally) are called. catch with { ... } This clause is invoked when an error happens in the "try" block, and the class of the error object satisfies the ErrorSelector specified. You may specify many "catch" clauses, to deal with different error situations. The can be any of the following forms: # Class matches a specific error class catch ErrorClass with { ... } # Class matches a specific regexp catch qr/.../ with { ... } # A subroutine returns a true value catch sub { ... } with { ... } # One of a list of error selectors catch selector1, selector2, selector3 with { ... } # All of an array list of selectors catch [ selector1, selector2, selector3 ] with { ... } NOTE: This is a major difference from Error.pm, which only allows a single class as a selector. except { ... } This clause is invoked when there is an error from the "try" block, but no "catch" clauses were invoked. otherwise { ... } This clause is invoked when no error occurs in the "try" block. finally { ... } This clause is always invoked as the final step in the "try" sequence, regardless of whatever things happen. throw("..."); The throw keyword is not actually exported. It is a method call on the Error object. So you can use it indirectly or directly. These two calls are identical: throw MyError("Something is wrong"); MyError->throw("Something is wrong"); You can also use throw to reraise an error in a catch/except block, like this: $@->throw(); assert($value, "assertion message"); This function will throw an AssertionError error unless $value is true. ERROR OBJECTS All errrors are Perl objects. They all have the 'Error' class as their topmost parent class. They all have the following methods and properties: throw Error($msg [, %properties]); This method throws a new instance of the error class. It is described more fully above. $@->text() The "text" method gets or sets the error message for the object. Stringification All Error objects turn into their "text" string value when used in string context. Numification All Error objects turn into a unique number when used in numeric context. PREDEFINED CLASSES The "errors" module defines a number of error classes that it uses to cast errors into. You can also create error objects yourself using these classes. The classes are defined in a hierarchy: + Error + StandardError + ArithmeticError + DivideByZeroError + AssertionError + IOError + IOFileError + IOFileOpenError + NotImplementedError + SyntaxError + RuntimeError + UserError + user defined errors should inherit from this Some of these are obvious. Some deserve elaboration. AssertionError Indicates a failed "assert" call. SyntaxError Indicates a bad string eval. NotImplementedError You can throw this in a stub subroutine. RuntimeError Indicates an unknown system error. NOTE: These error classes are still being determined. This list is not yet complete. The current hierarchy was influenced from these sources. * http://search.cpan.org/perldoc?autodie#CATEGORIES * http://www.python.org/dev/peps/pep-0348/#new-hierarchy FAQ Q: What is the difference between 'errors' and 'exceptions'? A: Four letters. Q: Is "errors" performant? A: Yes. Very. The module is small, simple, has no dependencies and no string evals. Q: Why another error/exception module? A: Because it has the perfect name. ACKNOWLEDGEMENTS The original code and ideas for this module were taken from Error.pm. AUTHOR Ingy döt Net COPYRIGHT Copyright (c) 2009. Ingy döt Net. This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself. See http://www.perl.com/perl/misc/Artistic.html POD ERRORS Hey! The above document had some coding errors, which are explained below: Around line 302: You forgot a '=back' before '=head1' You forgot a '=back' before '=head1'