Tuesday, July 16, 2013

How I got to my first finished test

   Sorry, I am afraid that I have neglected my blog as of late.  Life has a way of getting in the way!  I have had my plate full to the brim doing my OPW internship, and at the same time I am doing a lot of juggling coordinating my kids summer schedules and mine.  It has been an extremely hectic few weeks, but I hope at this point things are settling down, and I will have more opportunity to blog.

   As I mentioned in the last entry, my project this summer is to provide a battery of tests for the new WYSIWYG editor feature of Wikipedia, called Visual Editor.  Visual Editor is an important new feature that was launched in July to all logged-in and anonymous users as the default editor for the English-language Wikipedia. It is expected to be deployed to almost all Wikipedias by the end of July, making it available for all editorsThere are a huge number of things to be tested within Visual Editor, so I am planning to do a battery of tests on some parts of it, rather than attempt to cover it all.  One of the first things I decided to test is to make sure that links are working properly within the Visual Editor.  As I wrote the test case for the links feature, I was pleased that I also made progress towards my individual goal of learning Cucumber, Ruby and Page Object Design.  Finally, last week after 19 commits over several weeks,  I was able to reach a milestone and commit the test as finished.  After this last commit, my links test was merged into our master repository.  The link test is now part of the automated browser-level tests that run continuously on our Jenkins instance, doing compatibility and regression testing.

  It is a good feeling to know that I have completed a test that is already being run by WMF to check that links are functioning properly in the newly released Visual Editor feature.  So what did it take to get to this final commit for my Visual Editor links test?  Well, a lot of elbow grease, more than anything else.  This is what I've always found to be the main ingredient when coding.  A lot of trial and error has to happen before you get to the point where you are able to get your final code committed.  In  this case, there were 19 commits along the way to getting the test case finished.  There was also a lot of help I received along the way from my mentor Chris McMahon, who pair programmed with me several times as we developed the tests.  I have found pair programming is an extremely effective technique, and I'm pleased that Chris and Zeljko have introduced me to the concept.  I have found that you can learn much more efficiently and quickly when you pair program, especially at the earlier stage of coding in new languages.

  I ran into several other issues as I developed the code for the link test.  One early issue was that Chris realized some of my Cucumber code duplicated code we had already written for other tests, so he had to refactor the Cucumber links script I had written.  This meant that I had to redo some of the coding on my Ruby scripts that corresponded to changes he made to the Cucumber script after refactoring.  I believe that due to the fact we were pair programming, this was caught before too much time went by, and it was a relatively easy change for me.  If I had handed the test case over after weeks of working on it by myself, it probably would have taken me much longer to rework the code.  This is example of how pair programming can work well, and can end up saving time in the end.

  Another problem I ran into as I developed my Cucumber feature script was that the sequence of steps I had created in Cucumber did not create the proper link in the Link UI.  Upon looking at the Link UI, I realized I was missing a step.  I then had to add the missing step to my Cucumber script to make it work properly.  Another issue I encountered towards the end of my coding was the need to reformat some of my scripts to get rid of the extra white-space and new lines.  It was also necessary to rename the script from ve_links.feature to visual_editor.links.feature to correspond to the naming conventions we were using.  In a nutshell, these were some of the challenges I faced when creating my links test.

Now for some public service announcements concerning Wikimedia automation testing..

Coming this Thursday, July 18th...
 QA automation workshop at  16:00 UTC !  Workshop is being held at this time especially for our Asia/European volunteers to get onboard!  Please sign up at the link if you are interested in learning or working with Cucumber and Ruby and contributing to our test case coverage.



Can't make it but are interested in learning more about current and future WMF QA activities?  Please sign up at our Wikimedia QA mailing list.

Further questions or interested in volunteering to write tests?  Feel free to email me at rachelqa99 [at] gmail [dot] com .

Thank you for your interest :)

1 comment:

  1. I am glad you like pair programming, there is more to come! :)

    ReplyDelete