Skip to content

Latest commit

 

History

History
276 lines (168 loc) · 7.73 KB

TESTING.rst

File metadata and controls

276 lines (168 loc) · 7.73 KB

Testing

This document provides instructions and recipes for testing the code before submitting.

The only required procedure is as follows:

  1. Make sure, you have tox installed:

    $ pip install -U tox
    
  2. Run all tests across all locally available python versions:

    $ tox
    

Make sure, all tests are passing.

If it runs in at least one python version, you may ignore the ERROR: pyXY: InterpreterNotFound: pythonX.Y errors.

The testing infrastructure built on top of set of tools, explained in this section.

Test runner collects available tests, runs some or all of them, reports problems and even allows jumping into command line debugger if you need it.

There are multiple test runners available, here we use pytest.

pytest is mostly installed by running tox (see below), but can be installed by:

$ pip test pytest

After that, command py.test shall be available.

Tests are placed in tests subdirectory or subdirectories.

To run all tests in tests subdirectory, be a bit verbose and print possible output printed to stdout by your testing code:

$ py.test -sv tests

To run all tests defined in tests/test_it.py:

$ py.test -sv tests/test_it.py

pytest allows running most tests written for unittest, nose and other testing frameworks.

pytest also allows writing test cases using so called fixtures, which allow writing very readable modular test suites.

Another strong point of pytest is error reporting. Generated reports are very well readable.

Virtual environments, or "virtualenv" in short is a tool, allowing to run your python code in isolated python environment. It has the advantage, that you may safely install into it without spoiling global python environment or being restricted by what is installed in the global environment.

virtualenv can be set up by means of various tools (virtualenv, tox, virtualenvwrapper, venv and others).

In our case, we rely on tox command to create well defined virtual environment, which can be easily recreated, extended and activated.

For that reason, there is no reason to install virtualenv separately.

Building and testing python code might be complex as it involves building fresh version of tested package, installing it incl. dependencies, installing packages needed for testing, running test. This all is often needed in multiple python versions and preferably shall be done using virtual environments.

With tox, you just jump into directory with tox.ini and run:

$ tox

Test coverage is checking, what lines of tested code were executed during tests. The higher percentage the better.

The tool even allows to show checked code and highlight the lines, which were not executed. There are often the lines, where bugs are present.

This project is using coverage tool.

This section provides practical instructions for specific activities related to testing and development.

The command:

$ tox

Will run all tests in all relevant python interpreters installed on your machine.

Relevant interpreters are defined in tox.ini file.

In case some python version is to be tested but is not installed, it reports ERROR: pyXY: InterpreterNotFound: pythonX.Y error. This shall not happen for the version you are developing, for other versions you can safely ignore this message.

Warning

`tox` shall not be started in activated virtualenv. If you run tox in activated virtualenv, you may experience strange results.

First, check, what python versions is tox configured to run for:

$ tox -l
py26
py27
py33
py34

Assuming you want to run tests for python 2.7, run:

$ tox -e py27

Running tox, one or more virtualenvs are always created:

$ tox -e py27

Virtualenvs are by default located in .tox directory:

$ dir .tox
dist
log
py27

Note

Ignore the dist and log directories. There may be more directories like py34.

To activate virtualenv on Linux:

$ source .tox/py27/bin/activate

On MS Windows:

$ source .tox/py27/Scripts/activate

After you activate virtualenv, command prompt often stars showing name of the virtualenv, e.g.:

(py27) $

This is shell specific behaviour.

You may install new packages into activated environment:

$ pip install anotherpackage

To deactivate virtualenv:

$ deactivate

If you need to recreate the environment, you may either remove given directory, or ask tox to recreate it:

$ tox -e py27 -r

First, create and activate virtualenv for your target python version (as described above).

Then run all the tests:

$ py.test -sv tests

See pytest help for more options (see e.g. --pdb which starts command line debugger in case some test fails).

First, create and activate virtualenv for your target python version (as described above).

Then run the test of your interest:

$ py.test -sv tests/test_it.py

For more methods of selecting tests see: Specifying tests / selecting tests

First, create and activate virtualenv for your target python version (as described above).

To test coverage of complete package awslogs:

$ coverage run --source awslogs -m pytest tests/

This creates or updates a file .coverage.

To show coverage report on console:

$ coverage report
Name                    Stmts   Miss  Cover
-------------------------------------------
awslogs/__init__.py         2      0   100%
awslogs/bin.py             85      9    89%
awslogs/core.py           143     12    92%
awslogs/exceptions.py      12      2    83%
-------------------------------------------
TOTAL                     242     23    90%

You may also generate nice HTML report, showing your source code highlighting lines, which were ignored by your tests.

First create the HTML report:

$ coverage html

Then open it in your web browser, e.g. in Firefox:

$ firefox htmlcov/index.html

This task is relevant to package maintainer only.

When is all ready for release:

  1. Commit all changes to git
  2. Run the tests to be sure, all is really ready
  3. Add "version" tag to git, e.g.: $ git tag -a 9.10
  4. Build the distributable (more details below)
  5. Upload the distributable to pypi as usually

To build the distributable, there are few options.

Running tox creates the package in .tox/dist directory.

Or you may activate relevant virtualenv and run:

$ python setup.py <buld_command_you_prefer>