Find Your Failing Tests

When I start on a extract method refactoring, my normal trick is to raise in the method that I’m looking to extract. I then find all the test files that failed and stick them in a wip rake task. I can’t claim any credit for this particular method of doing things though. Both of those tricks came from a pairing session I had with Katrina Owen.

Previously, most of my refactoring work on Tracks was focused on smaller sections of the code and I hadn’t hit any large cross cutting concerns yet. This meant that I could just scan the test failures for the files in question, since they were rarely larger than a page or two.

The last refactoring I did, along with the current refactoring I’m working on, cut across various concerns and spew errors all across the test suite. Scanning was not going to work anymore. So I put together a quick shell command to find the failing test files for me:

rake test | grep ' test/' | cut -d':' -f1 | sort | uniq

This does the following:

1. Runs the tests
2. Looks for files that live in the test directory. The extra space was just to make sure I grabbed only the test files, since they’re indented by spaces in the Rake output.
3. Grabs the filename of the test. The test runner puts a colon between the filename and the line number, so we can split on that easily.
4. Sorts the files
5. Makes the list of files unique

At this point, I have a nice list of files with failing tests to plug in to my wip rake task and don’t have to run my whole test suite while I refactor. Enjoy!