Testing

Important

This document has been migrated from our old wiki as is, and has not yet been revised. The content might be outdated, links and images could be broken. We are aware and will fix any issues as soon as possible.

Testing Your Launchpad Changes

General usage

The normal pattern for testing your changes is to run the tests you think will be affected locally, but fundamentally we rely on post-merge testing by buildbot.

Iterating with testrepository

apt-get install testrepository
#cd $yourlaunchpaddevdir
testr init

Don’t worry about creating a .testr.conf file; the defaults created for you works fine.

To run all the tests:

testr run

To run an individual test using the `-t PATTERN` option:

testr run -- -t foo

To see the current known failures:

testr failing

To run just the known failing tests:

testr run --failing

To re-run the tests:

testr run --failing

To see the current failing tests

testr failing

testr is moving and bug reports and patches are accepted :).

Running old school

To run the tests, you run the

./bin/test

script, which is produced by

make build

You can see all the options you have by running

./bin/test --help

Usually you will run

./bin/test -vvct PATTERN

where

PATTERN

is a regular expression that is used to filter the tests that will be run.

You can use ‘!PATTERN’ to match all expression not matching that patter. Ex, to run all test, except the one at Windmill layer you can use:

./bin/test -vvc --layer '!Windmill'

Speed up the tests

Since Librarian and MemCache does not change often and they take a long time to be started and shutdown, ./bin/test can leave them started by using this command:

LP_PERSISTENT_TEST_SERVICES=1 ./bin/test PATTERN

You can kill them using

./bin/kill-test-services

When running tests written in python files, and you only want to test a file, you can speed up the test by specifying the full path to the python file:

LP_PERSISTENT_TEST_SERVICES=1 ./bin/test PATH/TO/PYTHON/TEST/FILE.py

See this mail for more.

Headless tests

  • If you are running browser tests on a machine without an X server, you can use xvfb-run:

xvfb-run -s '-screen 0 1024x768x24' bin/test YOUR_TEST_ARGUMENTS

Testing Launchpad Translations

  • Rosetta admin user - translations-deity@example.com

  • Pagetest browser setup for Rosetta Administrators rosetta_admin_browser = setupRosettaExpertBrowser()

  • Pagetest browser setup for Ubuntu Translation Administrators ubuntu_admin_browser = setupDTCBrowser()

Performance and stress tests

Populate the db

To add object into the database you can use:

& env LP_DBNAME="launchpad_dev" make iharness
from canonical.lp import initZopeless
zl = initZopeless()
#
# use the factory here
#
zl.commit()

Writing tests

For each part of an application there are 2 level of testing:

  • unit testing

    • comprehensive testing, including cornerstone cases

    • written in python files (try to avoid doctest)

  • smoke/functional/integration testing

    • testing normal use cases

    • written using doctest format

Model

  1. Unit testing

    • Files locate in lib/lp//scripts/tests

  2. Integration testing

    • Files locate in lib/lp//doc

View

  1. Unit/integration testing

    • Files locate in lib/lp//browser/tests

    • More details: ViewTests

  2. Smoke testing

    • Files locate in lib/lp//stories

    • More details: PageTests

Javascript

  1. Unit testing

    More details: TestingJavaScript JavascriptUnitTesting

  2. XHR integration testing

    • Use sparingly.

    • We use YUI tests with a full appserver behind it.

    • See standard_yuixhr_test_template.js and standard_yuixhr_test_template.py in the root of the Launchpad tree.

Scripts

  • Files locate in lib/lp//scripts/tests

API

  • Files located in lib/lp//stories/webservices

  • More details: TestingWebServices