Carp::Notify - Loudly complain in lots of places when things break badly

NAME

Carp::Notify - Loudly complain in lots of places when things break badly

VERSION

version 1.13

SYNOPSIS

Use it in place of die or croak, or warn or carp.

 # with Carp;
 use Carp;
 if ($something_a_little_bad) { carp("Oh no, a minor error!")};
 if ($something_bad) { croak ("Oh no an error!")};
 # with Carp::Notify;
 use Carp::Notify;
 if (something_a_little_bad) {notify("Oh no, a minor error!")};
 if ($something_bad) { explode ("Oh no an error!")};

DESCRIPTION

Carp::Notify is an error reporting module designed for applications that are running unsupervised (a CGI script, for example, or a cron job). If a program has an explosion, it terminates (same as die or croak or exit, depending on preference) and then emails someone with useful information about what caused the problem. Said information can also be logged to a file. If you want the program to tell you something about an error that's non fatal (disk size approaching full, but not quite there, for example), then you can have it notify you of the error but not terminate the program.

Defaults are set up within the module, but they can be overridden once the module is used, or as individual explosions take place. Please set up your appropriate defaults in the module. It'll save you headaches later.

IMPORTANT NOTE

This version is nearly identical to, and is bug-for-bug compatible with, version 1.10 which has been on CPAN for years but has not been indexed.

This public release is intended to catch the attention of anyone actually using this module and let them know that changes are coming!

REQUIRES

Perl 5.005, Socket (for emailing)

BUILT IN STUFF

Using the module
use Carp::Notify;

will require and import the module, same as always. What you decide to import it with is up to you. You can choose to import additional functions into your namespace, set up new default values, or give it a list of variables to store.

Carp::Notify will always export the explode function and the notify function. Carp always exports carp, croak, and confess, so I figure that I can get away with always exporting explode and notify. Nyaah.

Be sure that you set your default variables before using it!

make_storable
if this is an import argument, it will export the make_storable function into your namespace. See make_storable, below.

make_unstorable
If this is an import argument, it will export the make_unstorable function into your namespce. See make_unstorable, below.

croak
If this is an import argument, it will override the croak function in your namespace and alias it to explode. That way you can switch all of your croaks to explodes by just changing how you use your module, and not all of your code.

carp
If this is an import argument, it will override the carp function in your namespace and alias it to notify. That way you can switch all of your carps to notifies by just changing which module you use, and not all of your code.

(log_it|log_file|log_explode|explode_log|log_notify|notify_log|store_vars|stack_trace|store_env|email_it|email|return|subject|smtp|domain|port|die_to_stdout|die_quietly|die_everywhere|death_message|death_function|error_function)
Example: 
 use Carp::Notify (
        "log_it" => 1,
        "email_it => 0,
        "email" => 'thomasoniii@yahoo.com'
 );

These are hash keys that allow you to override the Carp::Notify's module defaults. These can also all be overloaded explicitly in your explode() calls, but this allows a global modification.

log_it
Flag that tells Carp::Notify whether or not to log the explosions to the main log file

log_file
Overrides the default log file. This file will be opened in append mode and the error message will be stored in it, if log_it is true.

If you'd like, you can give log_file a reference to a glob that is an open filehandle instead of a scalar containing the log name. This is most useful if you want to redirect your error log to STDERR or to a pipe to a program.

Be sure to use globrefs only explicitly in your call to explode, or to wrap the definition of the filehandle in a begin block before using the module. Otherwise, you'll be trying to log to a non-existent file handle and consequently won't log anything. That'd be bad.

log_file stores all notifications AND explosions

log_explode
Flag that tells Carp::Notify whether or not to log the explosions to the explosion log

explode_log
Overrides the default explosion log file. This file will be opened in append mode and the explosion message will be stored in it, if log_it is true.

If you'd like, you can give explode_log a reference to a glob that is an open filehandle instead of a scalar containing the log name. This is most useful if you want to redirect your explosion log to STDERR or to a pipe to a program.

Be sure to use globrefs only explicitly in your call to explode, or to wrap the definition of the filehandle in a begin block before using the module. Otherwise, you'll be trying to log to a non-existent file handle and consequently won't log anything. That'd be bad.

explode_log stores ONLY explosions

log_notify
Flag that tells Carp::Notify whether or not to log the notifications to the notification log

notify_log
Overrides the default notification log file. This file will be opened in append mode and the notification message will be stored in it, if log_it is true.

If you'd like, you can give notify_log a reference to a glob that is an open filehandle instead of a scalar containing the log name. This is most useful if you want to redirect your notification log to STDERR or to a pipe to a program.

Be sure to use globrefs only explicitly in your call to notify, or to wrap the definition of the filehandle in a begin block before using the module. Otherwise, you'll be trying to log to a non-existent file handle and consequently won't log anything. That'd be bad.

notify_log stores ONLY notifications

store_vars
Flag that tells Carp::Notify whether or not to list any storable variables in the error message. See storable variables, below.

stack_trace
Flag that tells Carp::Notify whether or not to do a call stack trace of every function call leading up to the explosion.

store_env
Flag that tells Carp::Notify whether or not to store the environment variables in the error message.

email_it
Flag that tells Carp::Notify whether or not to email a user to let them know something broke

email
Overrides the default email address. This is whoever the error message will get emailed to if email_it is true.

return
Overrides the default return address. This is whoever the error message will be coming from if email_it is true.

subject
Overrides the default subject. This is the subject of the email message if email_it is true.

smtp
Allows you to set a new SMTP relay for emailing

port
Allows you to set a new SMTP port for emailing

domain
Allows you to set a new SMTP domain for emailing

die_to_stdout
Allows you to terminate your program by displaying an error to STDOUT, not to STDERR.

die_quietly
Terminates your program without displaying any message to STDOUT or STDERR.

die_everywhere
Terminates your program and displays error message to STDOUT and STDERR.

death_message
The message that is printed to the appropriate location, unless we're dying quietly.

death_function
death_function can be called instead of death_message, for a more dynamic error message. It is handed a hash all of the Carp::Notify variables along with an additional key 'errors' containing the errors in the current explosion or notification.

death_function's use will cause the program to terminate with a simple 'exit'. You'll have to die within death_function yourself if you want to pass on that way.

death_function may be a coderef or a string containing a function name ('death_function', 'main::death', '&death', etc.)

error_function
error_function is called by Carp::Notify if it encounters an error (cannot open log file, or cannot connect to SMTP server, for instance), it is given one parameter, the literal error that occurred. You may set this parameter to do additional processing if Carp::Notify cannot complete error notification.

error_function may be a coderef or a string containing a function name ('error_function', 'main::error', '&error', etc.) =back

(storable variable)
A variable name within single quotes will tell the Carp::Notify module that you want to report the current value of that variable when the explosion occurs. Carp::Notify will report an error if you try to store a value that is undefined, if you had accidentally typed something in single quotes, for instance. For example, 
 use Carp::Notify ('$scalar', '@array');
 $scalar = "some_value";
 @array = qw(val1 val2 val3);
 explode("An error!");

will write out the values ``$scalar : some_value'' and ``@array : val1 val2 val3'' to the log file.

This can only be used to store global variables. Dynamic or lexical variables need to be explicitly placed in explode() calls.

You can store variables from other packages if you'd like:

use Carp::Notify ('$other_package::scalar', '@some::nested::package::array');

Only global scalars, arrays, and hashes may be stored in 1.00. 1.10 allows you to store a function call as well, the function will be called with the same arguements that death_message would receive. Its return value will be stored along with other stored variables.

make_storable
Makes whatever variables it's given storable. See storable variables, above. 
 make_storable('$new_scalar', '@different_array');

make_unstorable
Stops whatever variables it's given from being stored. See storable variables, above. 
 make_unstorable('$scalar', '@different_array');

explode
explode is where the magic is. It's exported into the calling package by default (no point in using this module if you're not gonna use this function, after all).

You can override your default values here (see Using the module above), if you'd like, and otherwise specify as many error messages as you'd like to show up in your logs.

 # override who the mail's going to, and the log file.
 explode("email" => "thomasoniii@yahoo.com", log_file => "/home/jim/jim_explosions.log", "A terrible error: $!");
 # Same thing, but with a globref to the same file
 open (LOG, ">>/home/jim/jim_explosions.log");
 explode("email" => "thomasoniii@yahoo.com", log_file => \*LOG, "A terrible error: $!");
 # don't log.
 explode ("log_it" => 0, "A terrible error: $!");
 # keep the defaults
 explode("A terrible error: $!", "And DBI said:  $DBI::errstr");

notify
notify is to explode as warn is to die. It does everything exactly the same way, but it won't terminate your program the way that an explode would.

Internal functions

error
Used to proxy to the registered error_function, if any.

today
Used to for standard formatting of the date.

simple_smtp_mailer
Used to speak to an SMTP server.

FAQ

So what's the point of this thing?

It's for programs that need to keep running and that need to be fixed quickly when they break.

But I like Carp

I like Carp too. :)

This isn't designed to replace Carp, it serves a different purpose. Carp will only tell you the line on which your error occurred. While this i helpful, it doesn't get your program running quicker and it doesn't help you to find an error that you're not aware of in a CGI script that you think is running perfectly.

Carp::Notify tells you ASAP when your program breaks, so you can inspect and correct it quicker. You're going to have less downtime and the end users will be happier with your program because there will be fewer bugs since you ironed them out quicker.

Wow. That was a real run-on sentence

Yeah, I know. That's why I'm a programmer and not an author. :)

What about CGI::Carp?

That's a bit of a gray area. Obviously, by its name, CGI::Carp seems designed for CGI scripts, whereas Carp::Notify is more obvious for anything (cron jobs, command line utilities, as well as CGIs).

Carp::Notify also can store more information with less interaction from the programmer. Plus it will email you, if you'd like to let you know that something bad happened.

As I understand it, CGI::Carp is a subset feature-wise of Carp::Notify. If CGI::Carp is working fine for you -- great, continue to use it. If you want more flexible error notification, then try out Carp::Notify.

But I can send email with CGI::Carp by opening up a pipe to send mail and using that as my error log. What do you have to say about that?

Good for you. I can too. But most people that I've interacted with either don't have the know-how to do that or just plain wouldn't have thought of it. Besides, it's still more of a hassle than just using Carp::Notify.

Why are your stored variables kept in an array instead of a hash? Hashes are quicker to delete from, after all

While it is definitely true that variables can be unstored a little quicker in a hash, I figured that stored variables will only rarely be unstored later. Arrays are quicker for storing and accessing the items later, since they're only accessed en masse. I'll live with the slight performance hit for the rarer case.

Can I store variables that are in another package from the one that called Carp::Notify?

You betcha. Just prepend the classpath to the variable name, same as you always have to to access variables not in your name space. If the variable is already in your name space (you imported it), you don't need the classpath since explode will just pick it up within your own namespace.

Can I store local or my variables?

Not in the use statement, but you can in an explicit explode.

Are there any bugs I should be aware of?

If you import explode into your package, then subclass it and export explode back out it won't correctly pick up your stored variables unless you fully qualified them with the class path ($package::variable instead of just $variable)

Solution? Don't re-export Carp::Notify. But you already knew that you should just re-use it in your subclass, right?

Always exports explode and notify no matter what is specified in the use statement. This is according to the original design but contrary to current conventions and is now considered to be a bug.

make_storable and make_unstorable don't scope the variables they store/unstore by the calling package, which means that they don't affect the results of store_vars.

store_vars destroys the contents of the stored variables, making multiple notifications from the same package not work well.

store_vars doesn't gracefully handle being called directly from the main script (i.e. it expects there to be at least one other function in the call stack).

The internal function today doesn't always report GMT offsets for some timezones that don't use whole hours (e.g. Indian/Cocos aka CCT, which is sometimes reported as +07-30 instead of +0630).

The internal function today has an edge condition where the GMT offset could report one additional or one less minute of offset (rare).

Could I see some more examples?

Sure, that's the next section.

Okay, you've convinced me. What other nifty modules have you distributed?

Mail::Bulkmail and Text::Flowchart.

Was that a shameless plug?

Why yes, yes it was.

Examples

 # store $data, do email the errors, and alias croak => explode
 use Carp::Notify ('$data', 'email_it' => 1, "croak");
 # email it to a different address, and don't log it.
 use Carp::Notify ("email" => 'thomasoniii@yahoo.com', 'log_it' => 0);
 # don't use global log file, log explosions and notifys separately.
 use Carp::Notify ('log_it' => 0, 'log_explode' => 1, 'log_notify => 1);
 # die with an explosion.
 explode("Ye gods!  An error!");
 # die with an explosion and a different email subject.
 explode ('subject' => "HOLY SHIT THIS IS BAD!", "The reactor core breached.");
 # explode, but do it quietly.
 explode ("die_quietly" => 1, "Ye gods!  An error!");
 # notify someone of a problem, but keep the program running
 notify ("Ye gods!  A little error!");

Version History

v1.12 - Mar 31, 2016 -

* Fix tests relying on hash ordering

* fix double adding of 1900 to year in calculations

* minor cleanups for Test::Perl::Critic and Test::Pod::Coverage

* eliminate conditional 'my' statements

v1.11 - Feb 1, 2013 -

* Updated documentation and maintainership.

v1.10 - April 13, 2001 -

* Long overdue re-write/enhancements.

* Cleaned up the internals.

* Added support for different log files for notifys/explodes.

* Added support for changing all the defaults externally.

* Improved today() function with Mail::Bulkmail's date generation.

* Added death_function to call upon dying instead of death_message.

* Added error_function to help handle unforeseen problems.

* Perl 5.6's 'caller' function returns an extra two fields of information. Jeff Boes pointed this out and extra-helpfully provided a patch for it as well.

* Added the ability to config the SMTP port. I was just being lazy in v1.00. :*)

v1.00 - August 10, 2000 - Changed the name from Explode to Carp::Notify. It's more descriptive and I don't create a new namespace.

v1.00 FC1 - June 9, 2000 - First publically available version.

AUTHORS

COPYRIGHT AND LICENSE

This software is copyright (c) 2016 by Brian Conry, and 2000, 2001 by Jim Thomason.

This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.

SUPPORT

Perldoc

You can find documentation for this module with the perldoc command.

  perldoc Carp::Notify

Websites

The following websites have more information about this module, and may be of help to you. As always, in addition to those websites please use your favorite search engine to discover more resources.

Bugs / Feature Requests

Please report any bugs or feature requests by email to bug-carp-notify at rt.cpan.org, or through the web interface at https://rt.cpan.org/Public/Bug/Report.html?Queue=Carp-Notify. You will be automatically notified of any progress on the request by the system.

Source Code

The code is open to the world, and available for you to hack on. Please feel free to browse it and play with it, or whatever. If you want to contribute patches, please send me a diff or prod me to pull from your repository :)

https://github.com/buc0/carp-notify

  git clone git://github.com/buc0/carp-notify.git
 Carp::Notify - Loudly complain in lots of places when things break badly