CGI::Log - Perl extension for centralized logging of debug, error, status and success messages from scripts or other modules.


  use CGI::Log;

  Log->debug("user: $user");            ## add messages
  Log->status("Welcome $user.");
  Log->error("I'm sorry $user, but you do not have access to that area.");

  @msg = Log->get_debug();              ## get messages
  @msg = Log->get_error();
  @msg = Log->get_error("UI");
  @msg = Log->get_status();

  Log->is_error;                        ## test for messages

  Log->debug_off;               ## causes print() and debug() to be skipped

  Log->print();                 ## outputs debug and error logs in HTML
  Log->clear;                   ## clear all entries (current pid)

  Log->_report;                 ## reports the sizes of the arrays (lengths)
  Log->ui_no_error();           ## turns off inclusion of $! in user error messages


This module acts as a central repository for debug, status and error messages. It instantiates itself automatically (if it needs to) so you can access the Log object functions from anywhere in you code including other modules/objects with a simple consistent syntax.

It was written for CGI and mod_perl programming, but it could easily be used in any perl script where there is a need for centralized logging. (The only function which is CGI specific is print() since it outputs the debug and error logs with HTML formatting.)

It was originally written to just hold debugging information, but it has been extended to hold information that you might want to return to the user (i.e. the user-interface).

It is designed to be very painless to use. Add the following to any script or module where you want to log messages:

        use CGI::Log;

The CGI::Log:: namespace has been aliased to Log:: in order to save a bit of typing when adding debugging messages. So, to add a debug message, enter:

        Log->debug("Your message here.");
        ## note: this is equivalent to CGI::Log->debug();

To add an error message:

        Log->error("Some information about the error goes here.");

To add an success, or status message:

        Log->status("A status or informational message for the user.");
        Log->success("Something worked properly.");

The following commands all retrieve the messages you've logged:

        @msg = Log->get_debug;
        @msg = Log->get_error;
        @msg = Log->get_error("UI");
        @msg = Log->get_status;
        @msg = Log->get_success;

Note: All the get_* methods return array references when called in scalar context. e.g.

        $msg = Log->get_success;        ## ref($msg) eq "ARRAY"

During CGI/mod_perl development it is very handy to dump all of the debugging messages at the bottom of the HTML page. This is done with:


This can just be left at the bottom of your main script. Logging can be turned off (default is on), and when it is turned off Log->print() doesn't do anything.

Types and Formats of Messages

Each of the four types of messages (debug, error, status, success) have slightly different logic. The differences are as follows:

The debug messages are not designed to be visable by the user (and in fact may be a security risk if you show the connection string for databases, etc.) The format of a debug message is:

        [caller:line [caller:line...]] message


        caller is a method name
        line is the line number from the method 
        message is your debugging message 

An example will make this clearer:


    1:      use CGI::Log;
    2:      &foo;
    3:      sub foo
    4:      {
    5:            Log->debug("We are on line 5 of the method: foo");
    6:            &bar;
    7:      }
    8:      sub bar
    9:      {
    10:           Log->debug("line 10 method: bar process id: " . $$);
    11:           Log->error("Error on line 11 in the method: bar");
    12:     }
    13:     Log->print();

When run it prints:

    -- DEBUG ( (pid: 3262) --
    [main:2 main::foo:5] We are on line 5 of the method: foo
    [main:2 main::foo:6 main::bar:10] line 10 method: bar process id: 456
    [ERROR] [main:2 main::foo:6 main::bar:11] This an error being called from the method: bar (No such file or directory)
    -- ERROR --
    [main:2 main::foo:6 main::bar:11] Error on line 11 in the method: bar (No such file or directory)

(Note: the HTML in the output has been removed for clarity.)

There are some things to note:

  • Each debug message includes the context of how it came to be called. This allows for your debug messages to be very short -- often just stating a simple fact such as "a is undefined" or showing the value of a variable.

  • Error messages are duplicated in the error and debug arrays, so that you can determine how your error message got called, and its relation to any debugging messages.

  • Error messages in the debug list have "[ERROR]" prepended to them.

  • Error messages include the contents of the error variable $! in brackets. This saves you from having to remember to include this variable in your error message.

Log messages of type "status" and "success" are not manipulated or modified. Whatever you put in is what you get back.

Log messages of type "error" are stored in two formats. The first is the format that in the output above. The second is suitable for returning to the user. (It doesn't include the call trace.) By default it includes the error message from the variable $!. If this is not desirable, call Log->ui_no_error()


  • It is nice to be able to add as many debugging messages without having to worry about slowing down your application when it gets deployed. Calling Log->debug_off will set the instance variable DEBUG_FLAG to undefined, and will prevent any messages in the current process from being stored. e.g.

            if ($config{DEBUG} eq "Off")    ## pretend %config holds global
            {                               ##   configuration info
            Log->debug("this debug message won't be saved because debugging is off.");
  • Even though the debug() function won't do anything when debugging is turned off (it returns immediately), there is still the overhead of a function call for each debug message. If you really want to get obsessive about performance you can try redefining the debug() method. e.g.

            use CGI::Log;
            sub CGI::Log::debug () { 1 };           ## redefine with prototype so it gets inlined
                                                    ## note: I haven't tested the efficiency of this!
                                                    ## note: prototypes are from perl 5.002 +
  • It is very handy to be able to turn on debugging from the URL. e.g.


    In your perl code you could have:

            if ($param{debug} ne "secret"           ## $param{debug} holds the CGI variable "debug" 
                    || $DEBUG != 1)                 ## $DEBUG is a config variable

    This can be a huge timesaver if access to the webserver is difficult. This can be huge trouble if you have confidential or security related information in your debugging messages. (That is why "debug" in the above example is the string "secret" and not "1" or something easy to guess.)

  • You can add nice status messages to your web application by doing something like:

            if (Log->is_error)
                    print "<font color=\"#ff0000\">ERROR</font><BR>\n";
                    for (Log->get_error) { print $_ . "<BR>\n"; }
            elsif (Log->is_success)
                    print "<img src=\"smiley_face.gif\">";
                    for (Log->get_success) { print $_ . "<BR>\n"; }
            for (Log->get_status) { print $_ . "<BR>\n"; }
  • Make sure you have "Log" and not "log" or you will get the run-time error:

            Can't take log of 0
  • At the end of your script (whether a CGI or mod_perl) you will almost always want a:


Documentation for CGI::Log was created by h2xs.


- too much noise in the debug call tracing under mod_perl. e.g.

        [main:0 (eval):0 Apache::Registry::handler:141 (eval):141 Apache::ROOT::perl::test_5flog_2epl::handler:16] debug message.

- not thread-safe.

- if you are using mod_perl and you do not remember to clean out the log with Log->clean(), you will waste lots of memory.

- CGI::Log takes the Log:: namespace by default. This might be seen as rude, or cause problems if it is already being used. (Check if %Log:: is defined???)


Jason Moore, 1998 <>