I'm trying to make a SOAP request from javascript using AJAX, thus getting my feet wet with two technologies and one programming language at the same time :^). I'm using a javascript SOAP client I've found here [guru4.net]. This client asks for the WSDL first, and does so by appending ?wsdl to the URL of the SOAP service.

So my simple SOAP service:

#!/usr/bin/perl

use warnings;
use strict; 
use lib '/home/hue/lang/perl/modules';  ## where Temperatures.pm lives
use SOAP::Transport::HTTP;

SOAP::Transport::HTTP::CGI->dispatch_to ('Temperatures')->handle;
[download]

Became:

#!/usr/bin/perl

use warnings;
use strict; 
use lib '/home/hue/lang/perl/modules';
use SOAP::Transport::HTTP;

use CGI;
my $q = CGI->new;
my @keywords = $q->keywords;

if (grep 'wsdl', @keywords) {
    print $q->header ('text/xml');
    open my $fd, '<', 'temp.wsdl' or die "open: $!";
    print while <$fd>;
    close $fd;
    exit;   
}

SOAP::Transport::HTTP::CGI->dispatch_to ('Temperatures')->handle;
[download]

This code has a problem: CGI->new eats the input to the script and chokes on it. I remedied it by doing my $q = CGI->new ('') but this prevents me from accesing $q->keywords. So I had to resort to $ENV{'QUERY_STRING'}. The final code reads:

#!/usr/bin/perl

use warnings;
use strict; 
use lib '/home/hue/lang/perl/modules';
use SOAP::Transport::HTTP;

use CGI;
my $q = CGI->new ('');  ## create empty query

if ('wsdl' eq $ENV{'QUERY_STRING'}) {  ## test for 'eq' is enough in t
+his simple case
    print $q->header ('text/xml');
    open my $fd, '<', 'temp.wsdl' or die "open: $!";
    print while <$fd>;
    close $fd;
    exit;   
}

SOAP::Transport::HTTP::CGI->dispatch_to ('Temperatures')->handle;
[download]

I realize that right now, CGI.pm is only used to send the proper header to the client. I think this is a good thing and see no point in removing CGI.pm from the equation, since probably Apache will have to load it for other scripts anyway. This runs under mod_perl, by the way.

Is this an acceptable way of doing it? Or is there a "better"^TM method?