Failing test for yielding label block

[seanpdoyle]

When yielding a block named label (via partial.label.yield), an argument error is raised.

Error:
RendererTest#test_render_partial_can_yield_block_named_label:
ArgumentError: wrong number of arguments (given 1, expected 2..4)
    test/renderer_test.rb:20:in `block (2 levels) in <class:RendererTest>'
    test/renderer_test.rb:19:in `block in <class:RendererTest>'

[seanpdoyle]

@kaspth I'm not sure what's at the root of this, but I encountered it by chance.

[seanpdoyle]

@kaspth the following diff passes the tests:

--- a/lib/nice_partials/partial.rb
+++ b/lib/nice_partials/partial.rb
@@ -95,11 +95,7 @@ module NicePartials
     end
 
     def method_missing(meth, *arguments, **keywords, &block)
-      if @view_context.respond_to?(meth)
-        @view_context.public_send(meth, *arguments, **keywords, &block)
-      else
-        define_accessor meth and public_send(meth, *arguments, **keywords, &block)
-      end
+      define_accessor meth and public_send(meth, *arguments, **keywords, &block)
     end
 
     def define_accessor(name)

It turns out that we were optimistically delegating to ActionView::Base whenever we could, instead of defining named sections.

Is this change too aggressive? Is there a gap in our test coverage of this use case, or are we safe to remove the delegation?

[kaspth]

Is this change too aggressive? Is there a gap in our test coverage of this use case, or are we safe to remove the delegation?

It's forpartial.helpers { def something = root_url }, so we can call view context methods without needing to be aware of it. I think we should add a test case for that behavior instead of removing the delegation.

I have an alternate idea where we can maybe remove the delegation need. Let me try that and submit a PR. I'll cc you on it.

[seanpdoyle]

@kaspth I've changed this PR's target branch.