Test Suite Design¶
The vast majority of the test suite is formed of HTML pages, which can be loaded in a browser and either programmatically provide a result or provide a set of steps to run the test and obtain the result.
The tests are, in general, short, cross-platform, and self-contained, and should be easy to run in any browser.
Most of the repository’s top-level directories hold tests for specific web
standards. For W3C specs, these directories
are typically named after the shortname of the spec (i.e. the name used for
snapshot publications under
/TR/); for WHATWG
specs, they are typically named after the subdomain
of the spec (i.e. trimming
.spec.whatwg.org from the URL); for other specs,
something deemed sensible is used. The
css/ directory contains test suites
for the CSS Working Group
Within the specification-specific directory there are two common ways of laying out tests: the first is a flat structure which is sometimes adopted for very short specifications; the alternative is a nested structure with each subdirectory corresponding to the id of a heading in the specification. The latter provides some implicit metadata about the part of a specification being tested according to its location in the filesystem, and is preferred for larger specifications.
For example, tests in HTML for “The History
are located in
Many directories also include a file named
META.yml. This file may define any
of the following properties:
spec- a link to the specification covered by the tests in the directory
suggested_reviewers- a list of GitHub account username belonging to people who are notified when pull requests modify files in the directory
Various resources that tests depend on are in
Tests in this project use a few different approaches to verify expected behavior. The tests can be classified based on the way they express expectations:
Rendering tests ensure that the browser graphically displays pages as expected. There are a few different ways this is done:
Reftests render two (or more) web pages and combine them with equality assertions about their rendering (e.g.,
B.htmlmust render identically), run either by the user switching between tabs/windows and trying to observe differences or through automated scripts.
Visual tests display a page where the result is determined either by a human looking at it or by comparing it with a saved screenshot for that user agent on that platform.
Manual tests rely on a human to run them and determine their result.