Skip to main content

Testing of integrated issues

Testing is an important part of the Moodle development process. Depending on the integration period (normal or continuous) testing can take place at different times. Your first priority as a tester should be to finish testing on the day they have been assigned. You should update the testing status so that the testing status is clear.

The testing process

note
  1. Tests are allocated during the week by the Integration Team. Tests are usually performed by a dedicated team of test engineers but sometimes they may be assigned to HQ developers too.
  2. HQ developers should check mail or search tracker to see which issues are assigned to them for testing.
  1. Pull latest integration from git://git.moodle.org/integration.git
  2. Test issue by following the Testing instructions
  3. Select Pass test or Fail test as appropriate, adding a short description of what was tested if not obvious
  4. If you find you cannot finish testing a particular issue, click Stop testing and let the Integration team know about it.
  5. When a test is failed the assignee is usually contacted and asked to respond or provide a fix. If the patch is provided late and there is no time to find a new tester then the issue will be reopened.
  6. Once the fail fix is integrated, it goes back to complete re-testing.
  7. It's the responsibility of the tester to test the issue again, unless the patch is from tester. If the tester provides fix patch then test will be re-assigned. changed.
  8. The tester who passes the issue will be set as the tester for the issue.
Failing a test session

For test sessions, if you encounter a failure, please fail the issue add a comment on the issue itself. If everything's good, add a comment in the session and complete the session. You may also comment on the issue and say that testing passes on your part.

Expectation from developer and peer-reviewer

Testing instructions are clear, concise, complete, and accurate. Where possible they should be easy to perform. Please follow the Testing instructions guide recommendations.

Expectations of the Integration team

  • Tests should be allocated when the issue is integrated.
  • The integration team may need to help/re-assign tests if developers are having problems.

Expectation from tester

  • Testing must happen always against up to date integration.git repository (unless testing instructions include some exceptional git operation). More specifically, testing must not happen against development branches for a number of reasons (based on old core stuff, missing interdependencies with other issues or changes performed along the integration process, upgrade problems...).
  • If tester is not available for testing, this should be raised ASAP.
  • Testers should try to finish testing as early as possible as they are assigned, so when tests fail, the issue assignee has as much time as possible to resolve it.
  • When a test fails, or new (related) regression found then fail test.
  • If tester is not sure of results or need explanation on testing instructions, then tester can either fail test with comments, or contact the assignee individually to raise the problem.
  • Testers should let the integration team know ASAP if they are facing any problems, need help or may not be able to complete their allocated tests
  • For any reason (big test, not enough time, not started testing yet), if a test is dragged to next day then the tester should leave comment on tracker, updating the status of testing and adding the expected time needed to finish testing.
  • When a test is passed, it is recommended to add some extra information that confirms that all works as expected. This could be a browser screenshot, terminal output...
  • All UI tests should be tested on currently supported themes.

Checking tests assigned to you

  1. Log in to Tracker.
  2. Visit Issues waiting to be tested page.

Differences in test process during continuous integration periods

During continuous integration the schedule is changed to allow faster iteration and for bug fixes to be applied more rapidly than the usual weekly cycle. The goal during this period is ... goal during this period is to release a new version of master multiple times per week. We try to keep the process more flexible during this time in order that developers who have less pressing issues than others can take the load off those concentrating on big fixes. It works best if we work together to help each other out.

danger

Priority is given to testing issues to ensure we can release regularly

Installing a local test site from the integration repository

Moodle uses two Git repositories for its source code. Their names are moodle.git and integration.git and they live at http://git.moodle.org. All submitted patches that were accepted during the integration review go to the integration.git first. Testers use integration.git as the source of the code to test. Patches that survive testing are then promoted to moodle.git and become the part of the official Moodle weekly build.

To obtain the code from the integration.git repository, follow the instructions at Git for Administrators page. But instead of cloning from git://github.com/moodle/moodle.git, use

git clone git://git.moodle.org/integration.git

as the very first command.

tip

Alternatively, you can also use Moodle Development Kit (MDK), and add the option --integration to install a Moodle instance based on integration.git.

Changing theme to another one

Ensure you have following setting in the config (it allows you to change the theme in the URL).

   $CFG->allowthemechangeonurl = true;

For changing to a theme named "yay" add ?theme=yay to the url.

Notes

  • If the issue requires an Oracle or MSSQL installation for testing, and you don't have one, docker may help you.
  • Any update should be added as a comment on the tracker issue being tested.
  • If testers pass or fail an issue by mistake, then please request the integration team to reopen it for testing.
  • Testers should not be involved in the bug fixing or review process.
  • If an issue cannot be fixed within a sprint and has to be reopened, the fix for sprint version should be removed and an appropriate backlog version set.