An experimental test framework that I’ve been tinkering with.
I’m using it to explore a few different aspects or potential approaches to testing:
- Setup vs. Action for Ruby tests
- Rerunning tests in Ruby
- Making it easier for developers to create their own application-specific testing constructs to better express the behaviour they are trying to test.
What I have is a reasonably-compact testing framework with some ‘modern’ features like:
- nesting contexts
- matcher syntax
- flexible runners including running tests on specific lines
- flexible reporting of test output
I have slightly stalled working on it because:
- In it’s most simple usage it’s not very different from RSpec
- I haven’t really pushed into any of the exploration of things like
actionand rerunning tests yet
- While I have a hunch that it has a nice API that others could use to build their own testing constructs, I haven’t really explored that either, and it’s quite possible that RSpec is just as easy to de/re-compose already.
Those of you who attend LRUG may have heard about the “test framework episode” that I had last year. Here’s a diagram I drew at the time; a crazy-wall of related test framework inventions and re-inventions:
I’ll try to remember what originally prompted it; I think I had been struggling to refactor a large suite of tests (using test-unit and shoulda), and I was getting frustrated with how many hoops it felt like we had to jump through to test concisely.
It’s related to thoughts I have had about Setup vs. Action for Ruby tests.
Stepping through each of the commits, you can see that I tried to grow the library only using concrete tests that provided examples of the behaviour I expected. I believe that the commits also demonstrate some of the key choices that a DSL implementer has to make; see Capturing behaviour in Ruby DSLs.