PLP - Perl in HTML pages


Lighttpd installation

lighttpd.conf configuration using mod_fastcgi|PLP::Backend::FastCGI:

    server.modules += ("mod_fastcgi")
    fastcgi.server += (".plp" => ((
        "bin-path" => "/usr/bin/perl -MPLP::Backend::FastCGI",
        "socket"   => "/tmp/fcgi-plp.socket",

Apache installation

httpd.conf for a mod_perl|PLP::Backend::Apache setup:

    <Files *.plp>
        SetHandler perl-script
        PerlHandler PLP::Backend::Apache
        PerlSendHeader On

Test script (test.plp)

        print "Hurrah, it works!<br>" for 1..10;


PLP is yet another Perl embedder, primarily for HTML documents. Unlike with other Perl embedders, there is no need to learn a meta-syntax or object model: one can just use the normal Perl constructs. PLP runs under FastCGI|PLP::Backend::FastCGI and mod_perl|PLP::Backend::Apache for speeds comparable to those of PHP, but can also be run as a standard CGI|PLP::Backend::CGI script.


See either CGI|PLP::Backend::CGI, FastCGI|PLP::Backend::FastCGI (recommended) or Apache|PLP::Backend::Apache. At least the following servers are supported:

  • Lighttpd
  • With mod_fastcgi|PLP::Backend::FastCGI or mod_cgi|PLP::Backend::CGI.

  • Apache
  • Either version 1 or 2. Using mod_fcgid, mod_fastcgi|PLP::Backend::FastCGI, mod_perl|PLP::Backend::Apache, or mod_action|PLP::Backend::CGI.

PLP Syntax

  • <: perl_code(); :>
  • With <: and :>, you can add Perl code to your document. This is what PLP is all about. All code outside of these tags is printed. It is possible to mix perl language constructs with normal HTML parts of the document:

        <: unless ($ENV{REMOTE_USER}) { :>
            You are not logged in.
        <: } :>

    :> always stops a code block, even when it is found in a string literal.

  • <:= $expression :>
  • Includes a dynamic expression in your document. The expression is evaluated in list context. Please note that the expression should not end a statement: avoid semi-colons. No whitespace may be between <: and the equal sign.

    foo <:= $bar :> $baz is like <: print 'foo ', $bar, ' $baz'; :>.

  • <(filename)>
  • Includes another file before the PLP code is executed. The file is included literally, so it shares lexical variables. Because this is a compile-time tag, it's fast, but you can't use a variable as the filename. You can create recursive includes, so beware! (PLP will catch simple recursion: the maximum depth is 128.) Whitespace in the filename is not ignored so <( foo.txt)> includes the file named  foo.txt, including the space in its name. A compile-time alternative is include(), which is described in PLP::Functions.

PLP Functions

These are described in PLP::Functions.

PLP Variables

  • The URI of the PLP document, without the query string. (Example: C)

    Used to be renamed to $ENV{PLP_NAME}, which is still provided but deprecated.

  • The filename of the PLP document. (Example: C)

    $ENV{PLP_SCRIPT} also still provided but deprecated.

  • The version of PLP.

  • Controls debugging output, and should be treated as a bitmask. The least significant bit (1) controls if run-time error messages are reported to the browser, the second bit (2) controls if headers are sent twice, so they get displayed in the browser. A value of 3 means both features are enabled. The default value is 1.

  • Contains a reference to the code that is used to report run-time errors. You can override this to have it in your own design, and you could even make it report errors by e-mail. The sub reference gets two arguments: the error message as plain text and the error message with special characters encoded with HTML entities.

  • %header, %cookie, %get, %post, %fields
  • These are described in PLP::Fields.

Things that you should know about

Not only syntax is important, you should also be aware of some other important features. Your script runs inside the package PLP::Script and shouldn't leave it. This is because when your script ends, all global variables in the PLP::Script package are destroyed, which is very important if you run a persistent backend (they would retain their values if they weren't explicitly destroyed).

Until your first output, you are printing to a tied filehandle PLPOUT. On first output, headers are sent to the browser and STDOUT is selected for efficiency. To set headers, you must assign to $header{ $header_name} before any output. This means the opening <: have to be the first characters in your document, without any whitespace in front of them. If you start output and try to set headers later, an error message will appear telling you on which line your output started. An alternative way of setting headers is using Perl's BEGIN blocks. BEGIN blocks are executed as soon as possible, before anything else.

Unless you're running as CGI, the interpreter won't exit after processing a page, so END { } blocks won't work properly. You should use PLP_END { }; instead. Note that this is a not a built-in construct, so it needs proper termination with a semi-colon (as do eval and do).

When run persistently, modules are loaded only once. A good modular design can improve performance because of this, but you will have to reload the modules yourself when there are newer versions.

The special hashes are tied hashes and do not always behave the way you expect, especially when mixed with modules that expect normal CGI environments, like Read PLP::Fields for information more about this.

FAQ and HowTo

A lot of questions are asked often, so before asking yours, please read the FAQ at PLP::FAQ. Some examples can be found at PLP::HowTo.


Currently maintained by Mischa POSLAWSKY

Originally by Juerd Waalboer


Copyright (c) 2000-2002 Juerd Waalboer, 2005-2008 Mischa POSLAWSKY. All rights reserved.

This software is free software; you can redistribute and/or modify it under the terms of the MIT/X11 license.


PLP::Functions, PLP::Fields, PLP::FAQ, PLP::HowTo