From: Jeremy Hinegardner <jeremy@...>
Date: 2008-03-08T01:48:51+09:00
Subject: Re: Net::HTTP.check_response very slow??

On Sat, Mar 08, 2008 at 01:39:13AM +0900, Tim Ferrell wrote:
> I am working on something that uses Net::HTTP to obtain the final url
> from a redirected url and, well, it seems really slow. I'm using ruby
> 1.8.6-p111 and have verified this behavior on OS X, Ubuntu Linux, and
> Windows XP.
> 
> A sample app is attached and here were my results, comparing the pure
> ruby approach to one shelling out to curl...
> 
> Tim@thinktank: tmp => time ./url_check.rb
> http://ia340903.us.archive.org/0/items/gd90-02-26.sbd.shakedown.451.sbeok.shnf/gd90-02-26d3t05_vbr.mp3
> 
> real  0m32.659s
> user  0m4.423s
> sys  0m3.168s

You are doing a full HTTP GET in this request and it is downloading the mp3.
Try:

  response = Net::HTTP.start('ia340903.us.archive.org') do |http|
                 http.head('/0/items/gd90-02-26.sbd.shakedown.451.sbeok.shnf/gd90-02-26d3t05_vbr.mp3')
             end

> Tim@thinktank: tmp => time ./url_check.rb --use-curl
> http://ia340903.us.archive.org/0/items/gd90-02-26.sbd.shakedown.451.sbeok.shnf/gd90-02-26d3t05_vbr.mp3
> 
> real  0m0.754s
> user  0m0.030s
> sys  0m0.022s

You are using curl's -I option here which odes an HTTP HEAD and only gets the
headers of the response, and does retrieve the 

> I expected the pure ruby approach to be slower but these results seem
> out of line just for a series of http response checks... Any ideas?
> 
> Cheers,
> Tim

enjoy,

-jeremy

-- 
========================================================================
 Jeremy Hinegardner                              jeremy@hinegardner.org