Coverage and Hotspot Analyzer Overview

The Coverage and Hotspot Analyzer is a tool for analyzing code coverage and hotspots in Perl programs and test suites. It can be used to test different versions or revisions of a program and compare them visually, or compare different test runs on the same code.

Each test run is referred to as an Experiment. Projects are logical groups of similar Experiments.

Experiments can be a simple command, the execution of a module's built-in test suite (e.g. `make test` on a CPAN module), or more complex testing scripts.

Experiments can be repeated using the same commands on new revisions of the code base, or repeated with some variation on the command or test script on the same code base. The results of these tests (runtime, code coverage, hotspots and other execution information) are stored in a database and can be easily viewed and compared.

The commands that produce experiment data are executed using perlcov-perl, a replacement perl interpreter which gathers the coverage and hotspot data and stores it in the Analyzer's database.

Typical Uses

Finding unexercised code

If you have a large program that uses several modules, you probably want to know if your test suite is actually running all the code. Running the test suite from within the Analyzer (e.g. `make test`) will show which parts of the code have not been run and which files have the least or most coverage.

Checking for changes in coverage

Using the Experiment model, it is easy to repeat the the same test suite on revised code to see changes in coverage. The coverage histogram gives an overview of the coverage for each experiment, and the Compare Coverage feature, lets you compare two experiments file by file and function by function.

Similarly, instead of changing the codebase, you could change the test suite and compare the coverage.

Finding hotspots

Hotspots are areas of code which are run more often than others, or that take longer to run than average. Using the View Hotspots and Compare Hotspots features it's easy to see and compare these. Hotspots can be viewed as Experiment Normalized, showing the percentage of the total experiment time used by a given line, or Function Normalized, showing the percentage of the subroutine execution time used by a given line.