Testing concurrent tasks / forking

Robbie Bow robbie at robbiebow.co.uk
Sun Dec 23 12:46:47 PST 2007


-------- Original Message  --------
Subject: Re:Testing concurrent tasks / forking
From: Tom Hukins <tom at eborcom.com>
To: Milton Keynes Perl Mongers <miltonkeynes-pm at pm.org>
Date: 23 December 2007 17:57:21
 > On Sun, Dec 23, 2007 at 01:05:20PM +0000, abhishek jain wrote:
 >> So you are interested in to running several concurrent requests at a
 >> time, i suggest u look at POE .
 >
 > I've never quite got my head round POE.  I find HTTP::Async
 > considerably easier to use.  But both these approaches might require
 > writing a fair amount of code, depending on what Robbie wants to do

That looks easier to use. I played with POE, trying to mix it with 
Test::WWW::Mechanize but am not convinced the end result was making 
concurrent connections. Here's an example of what I came up with: 
http://paste.husk.org/10587

The POE responses look to be coming back in the order they were sent, 
and take about the same time as the simple serial fetches. Even when I 
wrote a wee script to fetch that adds a nanosleep to each response and 
runs under FastCGI (see http://paste.husk.org/10586), the response all 
appear to come back in order sent.

HTTP::Async looks easier to use, but I'm not seeing a way to add CGI 
POST variables. I'll have a play anyway. My ideal situation will be to 
able to simulate several users submitting the same form to the web 
server at the same time, if that makes sense.

PS.http://www.robbiebow.co.uk/cgi-bin/sleepy.pl is dead. Long live 
http://www.robbiebow.co.uk/cgi-bin/sleepy.fcgi


More information about the MiltonKeynes-pm mailing list