difference between cgi and FCGI::Request in perl - perl

Question

I have a peice of code
my $fcgi_request = FCGI::Request(\*STDIN,\*STDOUT,\*STDERR, \%ENV,0,FCGI::FAIL_ACCEPT_ON_INTR);
while (1) {
my $cgi = new CGI('');
}
So what is the role of FCGI::Request and how it is different from CGI because new CGI() is also giving me request handle.

Answer 1

Disclaimer: This is mostly speculation.
It's likely that whoever implemented that code tried migrating an existing CGI application to FCGI to make it faster. If you look at the FCGI documentation you'll see that almost all the parameters that are passed to FCGI::Request are the default values anyway. The only not-default one is this (indentation mine, and actually in the code it's different).
flags (default: 0)
Possible values:
FCGI::FAIL_ACCEPT_ON_INTR
If set, Accept will fail if interrupted. It
not set, it will just keep on waiting.
Because those parameters are positional, the author of your code most likely thought it's a good idea to pass in all the defaults so not to change anything. That makes sense, but looks strange.
Then there is the loop in which the program accepts requests. Inside, there is the old CGI program. This program might use other features of CGI like HTML-generation (which is now discouraged), so the author probably left it in. Because of how CGI works (it reads environment variables) and the way FCGI works (also reads environment variables) that's not mutually exclusive.
Now if somewhere down in the program it says stuff like print $q->header('text/html') then that will go to STDOUT, which above was set as the filehandle that the FCGI request uses. So the FCGI handler will see the output, and all is well.
Now interesting is why there is a while (1) and not a while (my $fcgi_request->Accept() >= 0 ). It says that the Accept accepts a connection and attaches the handles. This should be in your code somewhere in the loop.

Answer 2

CGI is a protocol for passing an HTTP request to a process using a combination of environment variables and STDIN, and to receive a response using STDOUT. As such, the request can only be sent to a child created specifically to handle that request.
Launching a child for each request can be slow, especially if the program needs to be compiled before being executed. Fast CGI is a solution to this problem.
Fast CGI is is a protocol for passing an HTTP request to a process using a socket. As such, the request can be sent to a pre-existing process, and this process can handle multiple requests.
FCGI::Request accepts a Fast CGI request, and makes it look like a CGI request, allowing CGI.pm to be used. This allows familiar libraries to be used, and it allows CGI programs to be converted to Fast CGI programs with more ease.
In that particular snippet, FCGI::Request isn't actually used since $request->Accept() isn't being called. As written, it expects a CGI request, not a Fast CGI request.