Make click_failed_*.png world-readable

[eagletmt]

When we run tests using capybara-webkit in CI server, all users should
be able to read click_failed_*.png for debugging.

[mhoran]

Thanks for the pull request, @eagletmt. We're actually going to be removing this functionality, as capybara-webkit doesn't seem to be the right place to handle this. Check out #800 for further discussion. It may be worth kicking off an upstream discussion to figure out how to work this in.

[mhoran]

Note that a PR to do the same to the standard Render command would be welcome -- however we don't like to merge pull requests without tests.

[jferris]

I'm not sure it's a good idea for us to make decisions for users about who has access to their files. I think you can control how new files are created by changing your umask on CI.

[mhoran]

That's a good point. After commenting I noticed the files screenshots are created with world read on my machine. Sounds like there's a umask issue affecting the reporter.

[eagletmt]

It's not umask problem.
At least in Qt 5.4.2, QTemporaryFile seems to create a file with 0600 permission.
https://github.com/qtproject/qtbase/blob/5.4.2/src/corelib/io/qtemporaryfile.cpp#L176-L178

eagletmt% umask
022
eagletmt% git rev-parse HEAD
3ef6732baa2a3ed7a1b7ba5d6a464485779cebd4
eagletmt% bundle exec rspec spec/integration/session_spec.rb:308
Run options:
  include {:locations=>{"./spec/integration/session_spec.rb"=>[308]}}
  exclude {:full_description=>/Capybara::Session webkit Capybara::Window #(size|resize_to|maximize|close.*no_such_window_error)/, :skip_on_qt4=>false, :skip_if_offline=>false, :selenium_compatibility=>false, :skip_on_jruby=>false, :skip_on_windows=>false, :requires=>#<Proc:0x00000003ce9778>}

Capybara::Session
  iframe app
    raises an error if an element is obscured when clicked

Finished in 1.34 seconds
1 example, 0 failures
eagletmt% ls -l /tmp/click_failed_*
-rw------- 1 eagletmt eagletmt 17242 2015-07-02 23:39 /tmp/click_failed_J20297.png
-rw------- 1 eagletmt eagletmt 17242 2015-07-02 23:39 /tmp/click_failed_L20297.png
-rw------- 1 eagletmt eagletmt 17242 2015-07-02 23:39 /tmp/click_failed_M20297.png
-rw------- 1 eagletmt eagletmt 17242 2015-07-02 23:39 /tmp/click_failed_S20297.png
-rw------- 1 eagletmt eagletmt 17242 2015-07-02 23:39 /tmp/click_failed_T20297.png
-rw------- 1 eagletmt eagletmt 17242 2015-07-02 23:39 /tmp/click_failed_X20297.png
-rw------- 1 eagletmt eagletmt 17242 2015-07-02 23:39 /tmp/click_failed_h20297.png
-rw------- 1 eagletmt eagletmt 17242 2015-07-02 23:39 /tmp/click_failed_n20297.png
-rw------- 1 eagletmt eagletmt 17242 2015-07-02 23:39 /tmp/click_failed_o20297.png

[jferris]

@eagletmt interesting. I guess it makes sense for Qt to be generally paranoid about temporary files.

We only use QTemporaryFile in the automatic screenshot saving, though, which is being removed in #800. In the regular save_screenshot path, we use QImage.save, which seems to respect umask.

[eagletmt]

Okay, I understand this feature will be removed soon.