Fix undefined method `example_group' for nil & NoMethodError: render_views? for feature specs #1827
Conversation
…views? for feature specs
|
Maybe the author of #1732 @andymorris can also have a look at whether I preserved all behavior. As I said tests are green and to me it looks good. But I'm not deep into RSpec behavior. PR #1732 seemed to work for controller specs but not for feature specs. This PR here is trying to fix this. |
e.g. skip it for feature specs
|
ad40aae is only an optimization for the case when no controller specs are run. E.g. when only running feature specs. |
| end | ||
| end | ||
|
|
||
| LogSubscriber.attach_to(:action_view) |
There was a problem hiding this comment.
It's only an optimization for the case when no controller specs are run. E.g. when only running feature specs it should make things a little faster since not every template rendering is going through that Subscriber and so on. I could easily refrain from moving it.
| return if current_example_group.render_views? | ||
| info(" Template rendering was prevented by rspec-rails. Use `render_views` to verify rendered view contents if necessary.") | ||
| example_group = current_example_group | ||
| if example_group.respond_to?(:render_views?) && !example_group.render_views? |
There was a problem hiding this comment.
We prefer guard condition style returns, prehaps just change the first line to:
return if !example_group.respond_to?(:render_views?) || !example_group.render_views?
| RSpec.current_example.example_group | ||
| # When running feature specs current_example can be nil if the subscriber runs in a different context | ||
| # than the specs themselves. E.g. in a capybara thread. | ||
| RSpec.current_example.try!(:example_group) |
There was a problem hiding this comment.
Please don't use try! here, we try to avoid the usage of non standard Ruby, you could add a guard to return nil if theres no current example but really we should figure out why its nil.
There was a problem hiding this comment.
RSpec.current_example is nil in the case where the request came from the (Selenium) webdriver out of the browser (e.g. via AJAX) and not from an RSpec example group. I'll add that little extra info to the code comment.
There was a problem hiding this comment.
No don't thats not the cause, this is only executed during a test process when this should have a thread value, its not selenium causing it.
|
To mark Feature spec as Controller spec by :type => :controller like following could avoid this error message. Actually it's indeed Feature spec other than Controller spec, so I think this way is improper: describe "products page", :type => :controller do |
|
Fixed in #1831 |
|
(sorry I didn't see this PR until after I merged the other one, but I do think it addresses the feedback in here) |
This is my stab at issue #1800.
In my case the
ViewRenderercode that was added with PR #1732 was triggered via the capybara webdriver andRSpec.current_exampleis nil then causing the error in the title. @yujinakayama's spec repro in #1818 points to the cause being thatFeatureExampleGroupdoesn't includeRSpec::Rails::ViewRenderingwhileControllerExampleGroupdoes.This commit makes the tests in PR #1818 pass (via the
respond_to?) and running the whole rspec-rails suite only caused bundler errors for me (I tweaked the gemspec for local gem development). Using this local commit also got rid of loads of errors (those from the title) when running feature specs in my own project.I have to say that this is my very first experience with RSpec code and that I just deemed the
includedblock to be a good place to do this. Bear with me, if I forgot something.