I'm testing for deletion of records from an ActiveRecord class using Capybara/Selenium. The actual deletion is done using the Gherkin script (the xPath bit is so I can find the right user to delete):
When(/^I delete the other (.*)$/) do |role|
visit '/users'
page.evaluate_script('window.confirm = function() { return true; }')
find(:xpath, %{//tr[td='Bob']/td[@class='list_actions']//a[@data-method='delete']}).click
end
Unfortunately when I check the list of users, Bob is still there. However, if I pause execution (using an ask statement) then when I resume sure enough Bob has gone. So it seems that Capybara is checking for Bob's existence before Selenium/Rails has got around to deleting him.
Googling for this, there seems to be a lot of confusion over whether Selenium waits on a click action, and all of the solutions I can find assume something is being added so they can wait until it appears -- not the case in my scenario.
So how can I get rid of this race condition -- make Selenium wait until the record has been deleted?
