[rspec-users] Is there any way to use a second webrat session temporarily?

[Matt Wynne]

On 16 Mar 2009, at 23:35, Erik Pukinskis wrote:

> Dear Rspecucumberatubyists,
>
> I have a Rails app that creates subdomains, and I'd like to verify
> them in a cucumber feature like so:
>
>  When I visit http://blingo.myapp.com
>  Then I should see "Welcome to blingo!"
>
> The trouble is, webrat's visit method strips the domain off, and keeps
> only the path, so it ends up going to my site root instead of the
> subdomain.
>
> I tried creating a separate mechanize session that is used for full  
> urls:
>
> require 'webrat/mechanize'
>
> Given /^I visit (.*)$/ do |url|
>  if url.match(/^http:\/\//)
>    session = Webrat::MechanizeSession.new
>    session.visit(url)
>    response = session.response.body
>  else
>    visit url
>  end
> end
>
> And that seems to work, but then when my "Then" step goes to find the
> response, it doesn't find anything.  The "response" variable becomes
> nil somewhere along the way. The step that's matching there is the
> default one that ships with cucumber:
>
> Then /^I should see "([^"]*)"$/ do |text|
>  response.should contain(text)
> end
>
> Why is my response getting erased?
>
> Love,
> Erik

I think response is a method or read-only property, set by webrat when  
you call #visit, so it's not something you can just set like that. I  
would have a look in the webrat source code and see if you can figure  
out why the subdomain is being stripped out of the request. I have a  
feeling I remember someone asking about this before.

Also note that webrat has its own mailing list on google groups which  
might be a better place to find answers.

HTH,

Matt Wynne
http://blog.mattwynne.net
http://www.songkick.com