Rails3: How can I use UTF-8 when testing with Capybara

I use Rails 3, Capybara 0.4.0, Akephalos (https://github.com/sorentwo/akephalos) Selenium-webdriver and Rspec 2. I use ruby 1.9

I have the following spec:

  scenario "Create a new Group course with complex name" do
      school = School.make!
      switch_to_subdomain(school)
      course_group = CourseGroup.make

      click_link("Asignaturas")
      click_link("Nuevo grupo")
      fill_in('Nombre del curso', :with => 'Cuarto de E.S.O - Opción de Ciencias')
      fill_in('course_group_courses_attributes_0_name', :with => 'Laboratorio')
      save_and_open_page
      click_button("Create Course group")                     
      page.has_c开发者_如何学运维ontent?("Curso creado").should == true                                              
      course_groupDB = CourseGroup.where(:name => 'Cuarto de E.S.O - Opción de Ciencias').first
      course_groupDB.name.should == 'Cuarto de E.S.O - Opción de Ciencias'
    end  

If I use Selenium as a driver all is right. But When I use Akephalos I get:

Failure/Error: course_groupDB.name.should == 'Cuarto de E.S.O - Opción de Ciencias'
undefined method `name' for nil:NilClass

Which means that there is no model in the database with that name. It wasn't saved. The thing is, it was saved, but with non-utf characters.

How can I make sure that utf is used?

---

I made a fork that add support for utf8.

https://github.com/Nerian/akephalos