NAME
    CGI::SpeedyCGI - Speed up perl CGI scripts by running them persistently

SYNOPSIS
     #!/usr/bin/speedy

     ### Your CGI Script Here
     print "Content-type: text/html\n\nHello World!\n";

     ##
     ## Optionally, use the CGI::SpeedyCGI module for various things
     ##

     # Create a SpeedyCGI object
     use CGI::SpeedyCGI;
     my $sp = CGI::SpeedyCGI->new;

     # See if we are running under SpeedyCGI or not.
     print "Running under speedy=", $sp->i_am_speedy ? 'yes' : 'no', "\n";

     # Register a shutdown handler
     $sp->set_shutdown_handler(sub { do something here });

     # Register a cleanup handler
     $sp->register_cleanup(sub { do something here });

     # Set/get some SpeedyCGI options
     $sp->setopt('timeout', 30);
     print "maxruns=", $sp->getopt('maxruns'), "\n";

DESCRIPTION
    SpeedyCGI is a way to run CGI perl scripts persistently, which usually
    makes them run much more quickly. A script can be converted to SpeedyCGI
    by changing the interpreter line at the top of the script from:

        #!/usr/bin/perl

    to

        #!/usr/bin/speedy

    After the script is initially run, instead of exiting, SpeedyCGI keeps
    the perl interpreter running. During subsequent runs, this interpreter
    is used to handle new requests instead of starting a new perl
    interpreter for each execution.

    SpeedyCGI conforms to the CGI specification, and normally does not work
    inside the web server. A very fast cgi-bin, written in C, is executed
    for each request. This fast cgi-bin then contacts the persistent Perl
    process, which is usually already running, to do the work and return the
    results.

    Since the perl interpreter runs outside the web server, it can't cause
    problems for the web server itself. By default each perl script runs in
    its own Unix process, so one program can't interfere with another.
    Command line options can also be used to deal with programs that have
    memory leaks or other problems that might keep them from otherwise
    running persistently.

    SpeedyCGI also provides an Apache module so that under the Apache web
    server, scripts can be run without the overhead of doing a fork/exec for
    each request. With this module a small amount of frontend code is run
    within the web server - the perl interpreters still run outside the
    server.

OPTIONS
  Setting Option Values

    SpeedyCGI options can be set in several ways:

    Command Line
        The speedy command line is the same as for regular perl, with the
        exception that SpeedyCGI specific options can be passed in after a
        "--".

        For example the line:

                #!/usr/bin/speedy -w -- -t300

        at the top of your script will set the perl option "`-w'" and will
        pass the "`-t'" option to SpeedyCGI, setting the Timeout value to
        300 seconds.

    Environment
        Environment variables can be used to pass in options. This can only
        be done before the initial execution, not from within the script
        itself. The name of the environment variable is always SPEEDY_
        followed by the option name in upper-case. For example to set the
        speedy Timeout option, use the environment variable named
        SPEEDY_TIMEOUT.

    CGI::SpeedyCGI
        The CGI::SpeedyCGI module provides the setopt method to set options
        from within the perl script at runtime. There is also a getopt
        method to retrieve the current options. See the section on "METHODS"
        below.

    mod_speedycgi
        If you are using the optional Apache module, SpeedyCGI options can
        be set in the httpd.conf file. The name of the apache directive will
        always be Speedy followed by the option name. For example to set the
        speedy Timeout option, use the apache directive SpeedyTimeout.

  Context

    Not all options below are available in all contexts. The context for
    which each option is valid is listed on the "Context" line in the
    section below. There are three contexts:

    speedy
        The command-line "speedy" program, used normally with #! at the top
        of your script or from a shell prompt.

    mod_speedycgi
        The optional Apache mod_speedycgi module.

    module
        During perl execution via the CGI::SpeedyCGI module's getopt/setopt
        methods.

  Options Available

    BackendProg
            Command Line    : -p<string>
            Default Value   : "/usr/bin/speedy_backend"
            Context         : mod_speedycgi, speedy

            Description:

                Path to the speedy backend program.

    BufsizGet
            Command Line    : -B<number>
            Default Value   : 8192
            Context         : speedy

            Description:

                Use <number> bytes for the buffer that
                receives data from the CGI script.

    BufsizPost
            Command Line    : -b<number>
            Default Value   : 1024
            Context         : speedy

            Description:

                Use <number> bytes for the buffer that sends
                data to the CGI script.

    Group
            Command Line    : -g<string>
            Default Value   : "none"
            Context         : mod_speedycgi, speedy

            Description:

                Allow a single perl interpreter to run
                multiple scripts. All scripts that are run
                with the same group name and by the same user
                will be run by the same group of perl
                interpreters. If the group name is "none"
                then grouping is disabled and each
                interpreter will run one script. Different
                group names allow scripts to be separated
                into different groups. Name is
                case-sensitive, and only the first
                12-characters are significant. Specifying an
                empty group name is the same as specifying
                the group name "default" - this allows just
                specifying "-g" on the command line to turn
                on grouping.

    MaxBackends
            Command Line    : -M<number>
            Default Value   : 0 (no max)
            Context         : mod_speedycgi, speedy

            Description:

                If non-zero, limits the number of speedy
                backends running for this cgi script to
                <number>.

    MaxRuns
            Command Line    : -r<number>
            Default Value   : 500
            Context         : mod_speedycgi, module, speedy

            Description:

                Once the perl interpreter has run <number>
                times, re-exec the backend process.  Zero
                indicates no maximum.  This option is useful
                for processes that tend to consume resources
                over time.

    PerlArgs
            Command Line    : N/A
            Default Value   : ""
            Context         : mod_speedycgi

            Description:

                Command-line options to pass to the perl
                interpreter.

    Timeout
            Command Line    : -t<number>
            Default Value   : 3600 (one hour)
            Context         : mod_speedycgi, module, speedy

            Description:

                If no new requests have been received after
                <number> seconds, exit the persistent perl
                interpreter.  Zero indicates no timeout.

    TmpBase
            Command Line    : -T<string>
            Default Value   : "/tmp/speedy"
            Context         : mod_speedycgi, speedy

            Description:

                Use the given prefix for creating temporary
                files.  This must be a filename prefix, not a
                directory name.

    Version
            Command Line    : -v
            Context         : speedy

            Description:

                Print the SpeedyCGI version and exit.

METHODS
    The following methods are available in the CGI::SpeedyCGI module.

    new 
        Create a new CGI::SpeedyCGI object.

            my $sp = CGI::SpeedyCGI->new;

    register_cleanup($function_ref)
        Register a function that will be called at the end of each request,
        after your script finishes running, but before STDOUT and STDERR are
        closed. Multiple functions can be added by calling the method more
        than once. At the end of the request, each function will be called
        in the order in which it was registered.

            $sp->register_cleanup(\&cleanup_func);

    set_shutdown_handler($function_ref)
        Register a function that will be called right before the perl
        interpreter exits. This is not at the end of each request, it is
        when the perl interpreter decides to exit completely due to a
        Timeout or reaching MaxRuns.

            $sp->set_shutdown_handler(sub {$dbh->logout});

    i_am_speedy
        Returns a boolean telling whether this script is running under
        SpeedyCGI or not. A CGI script can run under regular perl, or under
        SpeedyCGI. This method allows the script to tell which environment
        it is in.

            $sp->i_am_speedy;

        To make your script as portable as possible, you can use the
        following test to make sure both the SpeedyCGI module is available
        and you are running under SpeedyCGI:

            if (eval {require CGI::SpeedyCGI} && CGI::SpeedyCGI->i_am_speedy) {
                Do something SpeedyCGI specific here...

    setopt($optname, $value)
        Set one of the SpeedyCGI options given in the section on "Options
        Available". Returns the option's previous value. $optname is case-
        insensitive.

            $sp->setopt('TIMEOUT', 300);

    getopt($optname)
        Return the current value of one of the SpeedyCGI options. $optname
        is case-insensitive.

            $sp->getopt('TIMEOUT');

INSTALLATION
    To install SpeedyCGI you will need to either download a binary package
    for your OS, or compile SpeedyCGI from source code. See the section on
    "DOWNLOADING" for information on where to obtain the source code and
    binaries.

  Binary Installation

    Once you have downloaded the binary package for your OS, you'll need to
    install it using the normal package tools for your OS. The commands to
    do that are:

    Linux
         rpm -i <filename>

    Solaris
         gunzip <filename>.gz
         pkgadd -d <filename>

    BSD pkg_add <filename>

    If you are also installing the apache module you will have to configure
    Apache as documented in the section on "Apache Configuration".

  Source Code Installation

    To compile SpeedyCGI you will need perl 5.004 or later, and a C
    compiler, preferably the same one that your perl distribution was
    compiled with. SpeedyCGI is known to work under Solaris, Redhat Linux,
    FreeBSD and OpenBSD. There may be problems with other OSes or other
    versions of Perl. SpeedyCGI may not work with threaded perl -- as of
    release 2.10, Linux and Solaris seem to work OK with threaded perl, but
    FreeBSD does not.

  Standard Install

    To do a standard install from source code, execute the following:

        perl Makefile.PL
        make
        make test
        make install

    This will install the speedy and speedy_backend binaries in the same
    directory where perl was installed, and the SpeedyCGI.pm module in the
    standard perl lib directory. It will also attempt to install the
    mod_speedycgi module if you have the command apxs in your path.

  Install in a Different Directory

    To install in a different directory, change the first line in the
    section on "Standard Install" to:

        perl Makefile.PL PREFIX=/somewhere

    This will install the binaries in /somewhere/bin and the SpeedyCGI.pm
    module under /somewhere/lib.

  Apache Installation

    To compile the optional apache mod_speedycgi module you must have the
    apxs command in your path. Redhat includes this command with the
    "apache-devel" RPM, though it may not work properly for installation.

    If the apache installation fails:

    *   Copy the mod_speedycgi.so from the mod_speedycgi directory to wherever
        your apache modules are stored (try /usr/lib/apache)

    *   Edit your httpd.conf (try /etc/httpd/conf/httpd.conf) and add the
        following lines. The path at the end of the LoadModule directive may
        be different in your installation -- look at other LoadModules to
        see.

            LoadModule speedycgi_module modules/mod_speedycgi.so
            AddModule mod_speedycgi.c

  Apache Configuration

    Once mod_speedycgi is installed, it has to be configured to be used for
    your perl scripts. There are two methods.

    Warning! The instructions below may compromise the security of your web
    site. The security risks associated with SpeedyCGI are similar to those
    of regular CGI. If you don't understand the security implications of the
    changes below then don't make them.

    1. Path Configuration
        This is similar to the way /cgi-bin works - everything under this
        path is handled by SpeedyCGI. Add the following lines near the top
        of your httpd.conf - this will cause all scripts in your cgi-bin
        directory to be handled by SpeedyCGI when they are accessed as
        /speedy/script-name.

            Alias /speedy/ /home/httpd/cgi-bin/
            <Location /speedy>
                SetHandler speedycgi-script
                Options ExecCGI
                allow from all
            </Location>

    2. File Extension Configuration
        This will make SpeedyCGI handle all files with a certain extension,
        similar to the way .cgi files work. Add the following lines near the
        top of your httpd.conf file - this will set up the file extension
        ".speedy" to be handled by SpeedyCGI.

            AddHandler speedycgi-script .speedy
            <Location />
                Options ExecCGI
            </Location>

BUGS / TODO
    Please report any bugs or requests for changes to speedycgi@newlug.org.
    The current bugs / todo list can be found at
    http://www.sourceforge.net/projects/speedycgi/. Go to the Bug Tracking
    menu and select the group "bug" for bugs, or the group "rfe" for the
    todo list.

FREQUENTLY ASKED QUESTIONS
    How does the speedy front end connect to the back end process?
        Via a Unix socket in /tmp. A queue is kept in a shared file in /tmp
        that holds an entry for each process. In that queue are the pids of
        the perl processes waiting for connections. The CGI frontend pulls a
        process out of this queue, connects to its socket, sends over the
        environment and argv, and then uses this socket for stdin/stdout to
        the perl process.

    If another request comes in while a CGI is running, does the client
    have to wait or is another process started?  Is there a way to set a limit
    on how many processes get started?
        If another request comes while all the perl processes are busy, then
        another perl process is started. Just like in CGI there is normally
        no limit on how many processes get started. But, the processes are
        only started when the load is so high that they're necessary. If the
        load goes down, the processes will die off due to inactivity, unless
        you disable the timeout.

        Starting in version 1.8.3 an option was added to limit the number of
        perl backends running. See MaxBackends in the section on "Options
        Available" above.

    How much of perl's state is kept when speedy starts another request?
    Do globals keep their values?  Are destructors run after the request?
        Globals keep their values. Nothing is destroyed after the request.
        STDIN/STDOUT/STDERR are closed -- other files are not. `%ENV',
        `@ARGV', and `%SIG' are the only globals changed between requests.

    How can I make sure speedy restarts when I edit a perl library used
    by the CGI?
        Do a touch on the main cgi file that is executed. The mtime on the
        main file is checked each time the front-end runs.

    Do I need to be root to install and/or run SpeedyCGI?
        No, root is not required.

    How can I determine if my perl app needs to be changed to work with
    speedy?  Or is there no modification necessary?
        You may have to make modifications.

        Globals retain their values between runs, which can be good for
        keeping persistent database handles for example, or bad if your code
        assumes they're undefined.

        Also, if you create global variables with "my", you shouldn't try to
        reference those variables from within a subroutine - you should pass
        them into the subroutine instead. Or better yet just declare global
        variables with "use vars" instead of "my" to avoid the problem
        altogether.

        Here's a good explanation of the problem - it's for mod_perl, but
        the same thing applies to speedycgi:

            http://perl.apache.org/faq/mod_perl_cgi.html#Variables_retain_their_value_fro

        If all else fails you can disable persistence by setting MaxRuns to
        1. The only benefit of this over normal perl is that speedy will
        pre-compile your script between requests.

    How do I keep a persistent connect to a database?
        Since globals retain their values between runs, the best way to do
        this is to store the connection in a global variable, then check on
        each run to see if that variable is already defined.

        For example, if your code has an "open_db_connection" subroutine
        that returns a database connection handle, you can use the code
        below to keep a persistent connection:

            use vars qw($dbh);
            unless (defined($dbh)) {
                $dbh = &open_db_connection;
            }

        This code will store a persistent database connection handle in the
        global variable "$dbh" and only initialize it the first time the
        code is run. During subsequent runs, the existing connection is re-
        used.

        You may also want to check the connection each time before using it,
        in case it is not working for some reason. So, assuming you have a
        subroutine named "db_connection_ok" that returns true if the db
        connection is working, you can use code like this:

            use vars qw($dbh);
            unless (defined($dbh) && &db_connection_ok($dbh)) {
                $dbh = &open_db_connection;
            }

MAILING LIST
    The mailing list address is speedycgi@newlug.org. Subscribe by sending a
    message to speedycgi-request@newlug.org with the word "subscribe" in the
    body.

    An archive of the mailing list is at
    http://newlug.org/mailArchive/speedycgi/ and mirrored at
    http://daemoninc.com/SpeedyCGI/mailArchive/

DOWNLOADING
  Binaries

    Binaries for many OSes can be found at:

        http://daemoninc.com/SpeedyCGI/CGI-SpeedyCGI/binaries

    There are also some older debian packages available from:

        http://www.debian.org/Packages/stable/interpreters/speedy-cgi-perl.html
        http://www.debian.org/Packages/unstable/web/libapache-mod-speedycgi.html

    The debian packages are not based on the latest version of SpeedyCGI.

  Source Code

    The standard source code distribution can be retrieved from any CPAN
    mirror or from:

        http://daemoninc.com/SpeedyCGI/download.html
        http://www.cpan.org/modules/by-authors/id/H/HO/HORROCKS/

    The latest development code can be obtained from the SourceForge CVS
    repository using the following commands:

     cvs -d:pserver:anonymous@cvs.SpeedyCGI.sourceforge.net:/cvsroot/SpeedyCGI login 
     cvs -z3 -d:pserver:anonymous@cvs.SpeedyCGI.sourceforge.net:/cvsroot/SpeedyCGI co 2.x

    Press Enter when prompted for a password.

AUTHOR
        Sam Horrocks
        Daemon Consulting Inc.
        http://daemoninc.com
        sam@daemoninc.com

SEE ALSO
    perl(1), httpd(8), apxs(8).

TRANSLATIONS
    Japanese
        http://member.nifty.ne.jp/hippo2000/perltips/CGI/SpeedyCGI.htm

COPYRIGHT
    Copyright (C) 2001 Daemon Consulting Inc.

    This program is free software; you can redistribute it and/or modify it
    under the terms of the GNU General Public License as published by the
    Free Software Foundation; either version 2 of the License, or (at your
    option) any later version.

    This program is distributed in the hope that it will be useful, but
    WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General
    Public License for more details.

    You should have received a copy of the GNU General Public License along
    with this program; if not, write to the Free Software Foundation, Inc.,
    59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.