09:17 <bradgonesurfing> I just started using mocks and they are wonderful. However I was wondering if it is possible to create mock expectations on an instance but still let those methods those expectations are applied against execute? I know this sounds heretical but I am just curious.

09:19 <technicalpickles> bradgonesurfing: that's called 'proxying' in the mock/stub world iirc. I don't recall offhand if rspec mocking supports it

09:20 <bradgonesurfing> technicalpickles: proxying. that was the word I was looking for. I'll go google a bit on it. Ta.

09:20 <technicalpickles> I know rr support it for example, but I don't seem to think rspec's mocking does

09:22 <technicalpickles> http://rspec.info/documentation/mocks/message_expectations.html -> Arbitrary Handling of Received Messages

09:22 <technicalpickles> I'm not sure if you can get acces to the original unstubbed method though

09:22 <technicalpickles> if you could, maybe you could call it in that block

09:24 <bradgonesurfing> technicalpickles: I could first alias the method and then call it through the block I guess.

09:25 <technicalpickles> yeah, but omghax :)

09:26 <bradgonesurfing> I know. Yuck but first hax and then monkey patch then happiness.

09:28 <technicalpickles> fair point

09:28 <technicalpickles> maybe that'd make sense to alias the method of stubbed stuff in general?

09:29 <technicalpickles> or maybe foo.should_receive(:blah).and_proxy !?!

09:36 <bradgonesurfing> I'm thinking foo.should_recieve!(:blah)

09:38 <technicalpickles> interesting take. I'm not sure it makes it obvious that it'd proxy to the original method though

09:38 <technicalpickles> should_proxy maybe?

09:48 <bradgonesurfing> technicalpickes: metaprogramming question. How to alias a method on an instance. Say I have object foo with method bar and I want to alias bar to sheep.

09:48 <bradgonesurfing> If I can do that then I think I can make the proxy.

09:54 <technicalpickles> bradgonesurfing: hmm... not sure offhand

09:54 <technicalpickles> you might be able to do it on an instance straight up. did you try yet?

11:02 <bradgonesurfing> technicalpickles: Ok here is the monkey patch and the spec to test it. It look ugly and my metaprogramming skills are not rock star quality but it seems to work. Any comments?

11:02 <bradgonesurfing> http://gist.github.com/337835

11:03 <technicalpickles> bradgonesurfing: first off, edit the gist to make the spec end in .rb or tell it otherwise tell it its ruby ;)

11:03 <bradgonesurfing> done

11:04 <technicalpickles> niice :)

11:04 <technicalpickles> I don't think you need to keep __rspec_sym__ around

11:04 <technicalpickles> it's just used in the class_eval

11:05 <technicalpickles> one good pattern I've seen in rails with the metaprogramming is to make comments with what the generated code might look like

11:05 <technicalpickles> ie fill in the vars for a particular example

11:07 <bradgonesurfing> How to remove a class var. @@__rspec_sym__ = nil gives the same effect but I don't think get's rid of it.

11:09 <technicalpickles> is that a rspec thing that was there?

11:09 <technicalpickles> I'm just saying I don't see it used in the class_eval where it isn't interpolated

11:11 <bradgonesurfing> No it is not used except in the interpolation. Couldn't figure out another way to get it in scope. Explicit set to nil at the end would clarify this.

11:14 <technicalpickles> I'm just not sure why it's a class var in the first place. couldn't you jsut interpolate #{sym}?

11:16 <bradgonesurfing> Doesn't work. Ruby scoping issue.

11:16 <technicalpickles> ohhh, I get it now, sorry

11:17 <bradgonesurfing> NameError in 'Foo is a monkey' undefined local variable or method `sym' for #>

11:17 <technicalpickles> right. I think you can probably do an instance_eval instead of class << self and class_eval

11:17 <technicalpickles> not sure if that'd help

11:18 <bradgonesurfing> alias only works on the class level.

11:18 <bradgonesurfing> so i think i have to open the singleton class.

11:18 <technicalpickles> drat

11:19 <bradgonesurfing> also it would be nice to use define_method and use a closure for the method but ruby 1.8 doesn't allow block parameters to closures so I can't properly forward the method call that way. So I am left with the hack of pushing in the class variable ( and hoping for no name collisions )

11:21 <bradgonesurfing> However I think the code is buggy because I have a more complex example that is failing.

11:32 <bradgonesurfing> technipickles: I figured out how to get rid of the class var. Much cleaner. http://gist.github.com/337835

11:33 <technicalpickles> haha, nice

11:33 <technicalpickles> oh metaclass, how we adore thee

11:34 <bradgonesurfing> Yeah.... I have to introduce a bunch of guys at work to ruby on monday. I think I'll steer clear of ( class<<self;self;end ) this

11:50 <bradgonesurfing> technicalpickles: If you are still there and still interested. The final working version. http://gist.github.com/337835 I needed a guard to stop multiple defining of the proxy method. Now the orderd matcher work as well. Would anybody want this in rspec?

11:52 <technicalpickles> bradgonesurfing: looks good

11:52 <technicalpickles> it's hard to tell :)

11:52 <technicalpickles> I'm sure it'd be useful as a gem at least. I did that for test spies

11:57 <bradgonesurfing> I might package it up later. For the moment I just tucked it away http://xtargets.heroku.com/2010/03/19/proxy-mocks-with-rspec/ for safe keeping. Thanks for the help and review.

11:59 <technicalpickles> bradgonesurfing: just use jeweler to make a quick gem ;) would make it easier for peeps to try it

11:59 <bradgonesurfing> You wrote that right ?

12:00 <bradgonesurfing> Jeweler that is?

12:00 <technicalpickles> one case I was thinking about is... what does the error look like when it fails? I suspect that because you use that placeholder @__rspec_obj__ ... it may say that object expected to have the method called on it, rather than the object your calling should_receive! on

12:00 <technicalpickles> guilty </plug>

12:00 <bradgonesurfing> :).. I think your concern is right. Hang on a tick let me check.

12:01 <bradgonesurfing> # expected :bar with (any args) twice, but received it once ./apricocentral/ruby/pspec/app_server_service_spec.rb:104:in `should_receive!'

12:01 <bradgonesurfing> whoops. sorry cut and paste foobar.

12:02 <bradgonesurfing> #<Object:0xb717f400> expected :bar with (any args) twice.

12:03 <bradgonesurfing> I guess I can override the class method on Object to return the wrong class. Evil ?

12:03 <technicalpickles> heh

12:04 <technicalpickles> I think this is good for a first pass. might want to look at how expectations get set with should_receive for the next pass though

12:05 <technicalpickles> anyone know how to add methods for particular types of specs? ie controller vs model vs integration?

12:07 <technicalpickles> ah-ha!

12:07 <technicalpickles> config.include(ControllerExampleHelpers, :type => :controller)

12:33 <bradgonesurfing> technicalpickles; I've got jeweler almost working but i get the error. "There was a problem saving your gem: Number is invalid" when it was trying to push to gemcutter ( I am an account )

12:33 <bradgonesurfing> ( I have an account ) I mean

12:34 <bradgonesurfing> missing version file. Working now.

12:39 <bradgonesurfing> technicalpickles: Your jeweler is nice. I now have a gem on gemcutter called proxymock. Thanks.

13:22 <technicalpickles> bradgonesurfing: sweeet

14:31 <BobFunk> is it just me or are there no specs for rspec-rails?

14:31 <BobFunk> found a bug and was looking for a way to add a spec and replicate it

15:55 <dchelimsky> BobFunk: which version of rspec-rails?

15:55 <dchelimsky> there are no specs right now for rspec-rails-2.x

15:55 <dchelimsky> we're basically smoke testing it by generating a bunch of specs and running them

15:56 <dchelimsky> but rspec-rails-1 has a comprehensive suite of specs

16:12 <BobFunk> rspec-rails-2.x

16:13 <BobFunk> found the rspec-dev repo now - is that the place to generate and run specs?

17:56 <GitHub106> rspec-core: master Rob Sanheim * 779d64c (1 files in 1 dirs): Clean up some describes names: ...

17:56 <GitHub106> rspec-core: master Rob Sanheim * 06c227e (2 files in 2 dirs): Bugfix: rspec shouldn't blow up if spec files don't match "_spec.rb" convention: ...

17:56 <GitHub106> rspec-core: master Rob Sanheim * 8e213d5 (3 files in 2 dirs): Format elasped time a little bit cleaner -- we do need exponential precision: ...

17:56 <GitHub106> rspec-core: master commits bb574fe...8e213d5 - http://bit.ly/c0nqQ2