Speedy RSpec

Faster testing by running only the needed tests.

Speedy RSpec computes which tests are relevant for certain code changes. It does this by tracing all calls during a test run and saving for each test all the files which are seen. At runtime it uses this file to compute which tests offer coverage for a particular file.

gem 'speedyrspec'

Set path for the trace_file. This is where SpeedyRspec stores its dependency graph.

SpeedyRspec.configure do |config|
  config.trace_file = 'speedy_traces.json'
end

In order to compute the smallest amount of tests needed to be run, the following steps should be followed.

1. Compute dependency graph. This stores in the trace_file the dependency list of each file. This process is best repeated weekly.

rake speedyrspec:collect

2. Run tests for the changed files.

2.1 Tests for app files.

rake speedyrspec:run <app/files>

2.2. Tests from git changes. Load list of changed files and compute which tests to run.

rake speedyrspec:run:git

SppedyRspec exposes the following rake tasks:

rake speedyrspec:collect   # run tests and collect trace information.
rake speedyrspec:run       # run tests that exercise code in files given as parameters.
rake speedyrspec:run:git   # run tests that exercise code in modiffied git files.
rake speedyrspec:show      # show which files should be run for given tests.
rake speedyrspec:show:git  # show which files should be run for modffied git files.

It is recommended to periodically collect traces. This can be done weekly in a test run followed by a commit of the traces file into your repository. This is possible with a configuration similar to the following:

SpeedyRspec.configure do |config|
  config.trace_file = 'speedy_traces.json'
end

Another option is to write the traces file in S3 bucket set-up for static hosting. At trace-collection the AWS access credentials will have to be stored in the environment variables AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY.