Hi Folks,
This is some code I wrote for testing "aggregated" or "bonded" ethernet
connections. It provides two methods, "split" and "join", for chopping up
HTTP::Request objects and gluing HTTP::Response objects together.
The namespace was chosen because this module uses the HTTP "Range" and
"Content-Range" headers.
I believe that others may find this module useful but I'd like to receive
comments before uploading it to CPAN.
Available immediately from:
http://kolea.ifa.hawaii.edu/~jhoblitt/pm/HTTP-Range-0.01.tar.gz
Cheers,
-J
--
NAME
HTTP::Range - Handle multi-segment HTTP requests
SYNOPSIS
require HTTP::Range;
require HTTP::Request;
require LWP::UserAgent;
require LWP::Parallel::UserAgent;
my $url = "http://example.com";;
my $segments = 2;
my $head_request = HTTP::Request->new( HEAD => $url );
my $head_response = LWP::UserAgent->new->request( $head_request );
my $get_request = HTTP::Request->new( GET => $head_request->uri );
# divide a single HTTP request object into many
my @requests = HTTP::Range->split(
request => $get_request,
length => $head_response->header( 'Content-Length' ),
segments => $segments,
);
my $pua = LWP::Parallel::UserAgent->new;
$pua->register( $_ ) foreach @requests;
my $entries = $pua->wait;
my @responses;
push( @responses, $entries->{ $_ }->response ) foreach keys %$entries;
# fuse many HTTP responses into a single object
my $res = HTTP::Range->join(
responses => [EMAIL PROTECTED],
length => $head_response->header( 'Content-Length' ),
segments => $segments,
);
print $res->as_string;
DESCRIPTION
This module provides class methods that allow you to divide an
HTTP::Request object into multiple smaller segments of the original
request and to fuse the HTTP::Response objects resulting from a seg-
mented transfer back into a complete resource. The segmentation is
accomplished via the use of the HTTP "Range" and "Content-Range" fields
as defined in "RFC2616".
This module aims to be useful for HTTP transfers across aggregated net-
work connections, as is possible for Ethernet with "IEEE 802.3ad". It
may also be advantageous on high latency network links where a single
TCP session won't scale to use all available bandwidth and it will
probably circumvent some HTTP bandwidth throttling techniques.
