Download the PHP package datto/phpunit-entropy without Composer
On this page you can find all versions of the php package datto/phpunit-entropy. It is possible to download/install these versions without Composer. Possible dependencies are resolved automatically.
Download datto/phpunit-entropy
More information about datto/phpunit-entropy
Files in datto/phpunit-entropy
Package phpunit-entropy
Short Description Provides tooling to allow
License MIT
Homepage http://datto.com
Informations about the package phpunit-entropy
Datto PHPUnit Entropy
Entropy is a library to provide tools for working with randomized testing in
PHPUnit as well as for testing systems that make use of PHP's random functions
such as rand
and shuffle
.
Use cases
Detect dependent tests
Sometimes unit tests require changes to global state, or cover functionality that alters that state. While this is never an ideal situation, this can often result in the creation of inadvertent dependencies on other tests.
Enabling Entropy's test shuffling functionality randomizes the order of your tests, helping to highlight these dependencies.
Entropy's test-shuffling also avoids changing the order of tests using the
@depends
functionality of PHPUnit; future work will work to ensure that
shuffling can still occur for these tests when appropriate.
Randomized testing
Sometimes you can find yourself in a situation where the range of inputs for a function is so large that taking a cross-section of those inputs is a more efficient use of your resources. But what happens if that cross-section has holes where your tests might fail?
One way to get around this is to use randomized inputs, where input is
generated in such a way as to get a random cross-section of the ranges. In
other words, using rand
or another such non-deterministic approach.
An extension of this, property-based testing is the practice of applying random input to your application or function, and observing that the output adheres to certain rules or has certain properties relative to the input, rather than precise testing of equality to a known result set.
By managing the random seed for your tests, Entropy allows not only the use of these approaches, but also to make them repeatable, either by accepting a fixed seed or recovering it from the last failed run until your test suite passed.
Including in your project
You can add this library to your project using Composer:
Running
Once configured, you can run your test suite as normal; the listener will look after itself. On test error or failure, the seed used will be stored in a temporary file, so that on the next run it will be reused, rather than a new seed generated.
Defining the seed
The seed for the random number generator is provided from up to four locations, presented here in priority order.
Environment variable
If set, the SEED
environment variable will be used to override any other
settings. It can be set via export
, but recommended use is to set it for the
current run only:
Configuration
The seed may be fixed by configuration; see the Configuration section below.
Last run
If a test run fails, the seed used is stored in a temporary file. This is then loaded on a subsequent test, and will persist until the suite succeeds again.
rand()
If no seed is set via the above methods, the final method is to use PHP's
rand
function.
Configuration
Entropy Test Listener
Once installed via composer, configuring the test listener is simply a matter
of altering your PHPUnit configuration file (often phpunit.xml
):
Arguments
seeding
enabled - boolean
If set to true, the random number generator will be seeded by the listener.
seed - integer
If you set the seed via this argument, only this value will be used to seed the random number generator. See the section on seed priority below.
file - string
If set, this file will be used to store the last failed random seed; it
defaults to [TMPDIR LOCATION]/phpunit-entropy-seed
.
shuffle - boolean
If set to true, the order in which unit tests are executed will be randomized
(except for suites where @depends
is in use). This is useful in determining
and identifying inter-test dependencies.
Future work
- Keep track of subsequent test runs
- Detect changes between runs
- Provide guiding output when tests are shuffled
Developer contact
Christopher Hoult [[email protected]](mailto://[email protected])