Home

Minitest::Heat helps you identify problems faster so you can more efficiently resolve test failures. It does this through a few different methods.

It collects failures and inspects backtraces to identify patterns and provide a heat map summary of the files and line numbers that most frequently appear to be the causes of issues.

It suppresses less critical issues like skips or slows when there are legitimate failures. It won't display information about slow tests unless all tests are passing (meaning no errors, failures, or skips)

It presents failures differently depending on the context of failure. For instance, it treats exceptions differently based on whether they arose directly from a test or from source code. It also treats extremely slow tests differently from moderately slow tests.

Markers get some nuance so that slow tests receive different markers than standard passing tests, and exception-triggered failures get different markers for source-code triggered exceptions (E) and test-triggered exceptions ('B' for 'Broken Test').

It also formats the failure details and backtraces to make them more scannable by emphasizing the project-relates lines from the backtrace.

It intelligently recognizes when an exception was raised from a test defintion vs. when an exception is genuinely triggered from the source code in order to help focus on fixing deeper exceptions first.

Failures are displayed ina fairly predictable manner but formatted to show the source code from the test so you can see the assertion that failed in addition to the summary of values that didn't satisfy the assertion.

Skipped tests are displayed in a simple manner as well so that it's easy to see the source of the skipped test as well as the reason it was skipped.

Slow tests get slightly more informative labels to indicate that they did pass, but they could use performance improvements. Tests that are particularly slow are called out with a little more emphasis so it's easier to focus on really slow tests first as they frequently represent the most potential for performance gains.

It also always displays the most significant issues at the bottom of the list in order to reduce the need to scroll up through the test failures. As you fix issues, the list becomes shorter, and the less significant issues will make there way to the bottom and be visible without scrolling.

For some additional insight about priorities and how it works, this Twitter thread is currently the best place to start.