strongqa / howitzer Goto Github PK

Use auto documentation from rdoc_documentation branch and up to date it

There is some interesting article http://www.elabs.se/blog/53-why-wait_until-was-removed-from-capybara

It is required to disable gems on rubygems

Rspec, Cucumber, etc.

May be Selenium is stable now, we need check it. Also, we need redesign locator_store

Remember, we should not generate real data here in execute_generator(options) method. Use Rspec Mock instead for RubiGen::Scripts::Generate.new.run method execution

cucumber 1.3.6 ⇢ 1.3.7
multi_json 1.7.9 ⇢ 1.8.0
poltergeist 1.3.0 ⇢ 1.4.0 ⇢ 1.4.1
rawler 0.1.7 ⇢ 0.1.8

Use RSpec Mock

Sometimes there can be case when current url page matches to more than one URL_PATTERN. It is good idea to implement validation like it is in Cucumber.

User RSpec mock

Now, if page is unexpected we have RuntimeError. it is nice to have something like IncorrectPageError instead.

Example:
RuntimeError: Current url: http://rdv4.reportdrop.net/admin/accounts, expected: (?-mix:a/\w+)
./pages/accounts_page.rb:30:in `launch_account'

See lib/vendor

Use auto documentation from rdoc_documentation branch and up to date it

Add validator that will be something like that:
validates 'error', type: 'PHP'
Cover new feature by tests

This issues assumes test coverage only *_generator.rb files, and correct copying appropriate files(compare template files with copied ones by bits). Please do not cover ruby code of template files by tests!

For now, there are tests for file, but tests for STDOUT and STDERR are still required

We have to create GETTING_STARTED.md file like https://github.com/thoughtbot/factory_girl/blob/master/GETTING_STARTED.md

It will allows end users to start quickly with Howitzer.

For now we use following:

class HomePage < WebPage
   URL_PATTERN = /#{Regexp.escape(settings.app_host)}\/?/
end

Where URL_PATTERN is used to make sure it's right opened page.

class WebPage
  def self.given
    new
  end

  def wait_for_url(expected_url, timeout=settings.timeout_small)
    end_time = ::Time.now + timeout
    until ::Time.now > end_time
      operator = expected_url.is_a?(Regexp) ? :=~ : :==
      return true if current_url.send(operator, expected_url).tap{|res| sleep 1 unless res}
    end
    log.error IncorrectPageError, "Current url: #{current_url}, expected:  #{expected_url}"
  end

  private
  def initialize
    wait_for_url(self.class::URL_PATTERN)
  end
end

Sometimes there are some kind of web sites, which use POST instead of GET. As result, all pages have "/" url. We have to implement general interface to validate page by different criteria. For instance, by Url, Title, element presence.

For better design it is required

We need to have some rails app, which will be not changed. And use howitzer for testing

We have to create GETTING_STARTED.md file like https://github.com/thoughtbot/factory_girl/blob/master/GETTING_STARTED.md

It will allows end users to start quickly with Howitzer.

It is required to implement flexible settings for next ones:

log_dir
html_log
required_clean_logs

For instance, testingbot-integration-fix3

Makes sense to generate warning error to console if some setting was deprecated and does not make sense anymore

Use auto documentation from rdoc_documentation branch and up to date it

To save memory, and speedup tests, do all pages and emails as singleton classes

Rspec integration and supporting Firebug + Firepath were added, please update documentation

Please use RSpec Mock to test right ordering method execution, times execution and argument execution