(CGI) A standard
for running external programs
from a World-Wide Web HTTP server
. CGI specifies how to pass arguments
to the program as part of the HTTP request. It also defines a set of environment variables that are made available to the program. The program generates output, typically HTML
, which the web server processes and passes back to the browser
. Alternatively, the program can request URL redirection
. CGI allows the returned output to depend in any arbitrary way on the request.
The CGI program can, for example, access information in a database
and format the results as HTML. The program can access any data that a normal application program can, however the facilities available to CGI programs are usually limited for security reasons.
Although CGI programs can be compiled programs, they are more often written in a (semi) interpreted language
such as Perl
, or as Unix
shell scripts, hence the common name "CGI script".
Here is a trivial CGI script written in Perl. (It requires the "CGI" module available from CPAN
#!/usr/bin/perl use CGI qw(:standard);
print header, start_html, h1("CGI Test"), "Your IP address is: ", remote_host(), end_html;
When run it produces an HTTP
header and then a simple HTML page containing the IP address
of the machine that generated the initial request. If run from a command prompt it outputs:
Content-Type: text/html; charset=ISO-8859-1 head title
Untitled Document /head body h1
CGI TestYour IP address is: localhost /body /html
The CGI program might be saved as the file "test.pl" in the appropriate directory on a web server, e.g. "/home/httpd/test".
Accessing the appropriate URL
, e.g. http://acme.com/test/test.pl, would cause the program to run and a custom page produced and returned.
Early web servers required all CGI programs to be installed in one directory called cgi-bin but it is better to keep them with the HTML files to which they relate unless they are truly global to the site. Similarly, it is neither necessary nor desirable for all CGI programs to have the extension ".cgi".
Each CGI request is handled by a new process. If the process fails to terminate for some reason, or if requests are received faster than the server can respond to them, the server may become swamped with processes. In order to improve performance, Netscape
developed the ISAPI
standard which allow CGI-like tasks to run as part of the main server process, thus avoiding the overhead of creating a new process to handle each CGI invocation. Other solutions include mod_perl and FastCGI.
Latest version: CGI/1.1.