Restful Authentication fails its tests

[Chuck_Remes]

Tonight I installed the 'restful authentication' plugin. The first thing I always do, after following the 'setup' instructions from the generator, is run the plugin's tests. Plus, I copied the proper 'include' line from the session_controller.rb to application_controller.rb.

The plugin failed several tests on its session_controller. Has anyone else seen this?

cremes$ ruby sessions_controller_test.rb
Loaded suite sessions_controller_test
Started
.FFF.....
Finished in 0.198387 seconds.

  1) Failure:
test_should_fail_cookie_login(SessionsControllerTest) [sessions_controller_test.rb:74]:
<false> is not true.

  2) Failure:
test_should_fail_expired_cookie_login(SessionsControllerTest) [sessions_controller_test.rb:67]:
<false> is not true.

  3) Failure:
test_should_fail_login_and_not_redirect(SessionsControllerTest)
    [/Users/cremes/Documents/development/ruby/library/config/../vendor/rails/actionpack/lib/action_controller/assertions/response_assertions.rb:26:in `assert_response'
     /Users/cremes/Documents/development/ruby/library/config/../vendor/rails/actionpack/lib/action_controller/assertions/response_assertions.rb:18:in `assert_response'
     sessions_controller_test.rb:29:in `test_should_fail_login_and_not_redirect']:
Expected response to be a <:success>, but was <302>

9 tests, 12 assertions, 3 failures, 0 errors

I confirmed that it is allowing logins where the password is clearly incorrect.

Also, I ran 'rake tmp:clear' to clear out all session files and other cruft.

This is on OSX 10.4.8 with Ruby 1.8.4 running Rails 1.2RC1 and the latest restful authentication plugin from its home repository.

cr

[treybean (treybean@gmail.com)]

That's funny. I just did the same thing.

It seems to me that logged_in is returning the symbol :false instead of
false. On my system, :false evaluates to true.

treybean$ irb
irb(main):001:0> a = :false
=> :false
irb(main):002:0> a
=> :false
irb(main):003:0> if a
irb(main):004:1> puts "a must be true"
irb(main):005:1> else
irb(main):006:1* puts "a must be false"
irb(main):007:1> end
a must be true
=> nil
irb(main):008:0>

I wonder if this is a recent change, or possibly something that might
be fixed my upgrading to Ruby 1.8.5

I'll post if I figure anything else out.

Trey

[Chuck_Remes]

That's funny. I just did the same thing.

It seems to me that logged_in is returning the symbol :false instead of
false. On my system, :false evaluates to true.

Yes, I just tracked this down myself. All over authenticated_system.rb the author uses the symbol :false when I think he means the value false.

For example:

    def logged_in?
      current_user ||= :false
    end

Interestingly, changing every occurrence of :false to false doesn't magically make all the tests complete successfully. Tests that used to work now fail and the tests that failed now work.

I wonder if this is a recent change, or possibly something that might
be fixed my upgrading to Ruby 1.8.5

I hope not!

[treybean (treybean@gmail.com)]

I definitely think Rick meant to use the symbol. After all, he even
uses this as a condition in current_user=. I looked at the svn log and
see that he made these changes yesterday. Regression maybe? My bet is
still on the version of ruby.

Trey

[Chuck_Remes]

Any suggestions on a stable tag for restful_authentication? I'd like to give a stable version of it a test drive.

cr

[Rick_Olson]

I use :false, because if I used false or nil it would hit the db again
after each hit. If someone wants to submit a patch with fixed tests,
that'd be great. I'm using this code in 2 apps now and it works fine.
I think there are some instances where it's expecting current_user to
return nil/false. Just use logged_in? instead.

[Rick_Olson]

Oh, apparently != is not the same as ||= . Silly mistake, it's fixed now.

http://bs.techno-weenie.net/!revision/2508

[Chuck_Remes]

Rev 2508 works nicely. All tests pass. Thanks for the quick fix.

cr