General Test Guidelines¶
File Paths and Names¶
When choosing where in the directory structure to put any new tests, try to follow the structure of existing tests for that specification; if there are no existing tests, it is generally recommended to create subdirectories for each section.
Due to path length limitations on Windows, test paths must be less that 150 characters relative to the test root directory (this gives vendors just over 100 characters for their own paths when running in automation).
File names should generally be somewhat descriptive of what is being
tested; very generic names like
001.html are discouraged. A common
test-topic is a short
identifier that describes the test. It should avoid conjunctions,
articles, and prepositions as it should be as concise as possible. The
integer that follows is normally just increased incrementally, and
padded to three digits. (If you’d end up with more than 999 tests,
test-topic is probably too broad!)
The test filename is significant in enabling specific optional features, such as HTTPS or server-side substitution. See the documentation on file names flags for more details.
In the css directory, the file names should be unique within the whole css/ directory, regardless of where they are in the directory structure.
By default, tests are served over plain HTTP. If a test requires HTTPS
it must be given a filename containing
test-secure.https.html, or be the generated service worker test of a
.any test. For more details see the documentation on
If a test must be served from an HTTP/2 server, it must be given a
Various support files are available in in the directories named
/css/support/. Reusing existing resources is encouraged where
possible, as is adding generally-useful files to these common areas rather than
to specific test suites.
Sometimes you may want to add a script to the repository that’s meant
to be used from the command line, not from a browser (e.g., a script
for generating test files). If you want to ensure (e.g., for security
reasons) that such scripts will only be usable from the command line
but won’t be handled by the HTTP server then place them in a
subdirectory at the appropriate level—the server will then return a
404 if they are requested.
For example, if you wanted to add a script for use with tests in the
notifications directory, create the
subdirectory and put your script there.
Tests are generally formatted as HTML (including XHTML) or XML (including SVG). Some test types support other formats:
WebDriver specification tests are expressed as Python files
The best way to determine how to format a new test is to look at how similar tests have been formatted. You can also ask for advice in the project’s matrix channel.
Except when specifically testing encoding, files must be encoded in
UTF-8. In file formats where UTF-8 is not the default encoding, they
must contain metadata to mark them as such (e.g.,
<meta charset=utf-8> in HTML files) or be pure ASCII.
Server Side Support¶
The custom web server supports a variety of features useful for testing browsers, including (but not limited to!) support for writing out appropriate domains and custom (per-file and per-directory) HTTP headers.
Tests should be as short as possible. For reftests in particular scrollbars at 800×600px window size must be avoided unless scrolling behavior is specifically being tested. For all tests extraneous elements on the page should be avoided so it is clear what is part of the test (for a typical testharness test, the only content on the page will be rendered by the harness itself).
Tests should generally avoid depending on edge case behavior of features that they don’t explicitly intend on testing. For example, except where testing parsing, tests should contain no parse errors.
This is not, however, to discourage testing of edge cases or interactions between multiple features; such tests are an essential part of ensuring interoperability of the web platform. When possible, use the canonical support libraries provided by features; for more information, see the documentation on testing interactions between features.
Tests should pass when the feature under test exposes the expected behavior, and they should fail when the feature under test is not implemented or is implemented incorrectly. Tests should not rely on unrelated features if doing so causes failures in the latest stable release of Apple Safari, Google Chrome, or Mozilla Firefox. They should, therefore, not rely on any features aside from the one under test unless they are supported in all three browsers.
Existing tests can be used as a guide to identify acceptable features. For language features that are not used in existing tests, community-maintained projects such as the ECMAScript compatibility tables and caniuse.com provide an overview of basic feature support across the browsers listed above.
and the files located in the directory named
common), only use language
features that have been supported by each of the major browser engines above
for over a year. This practice avoids introducing test failures for consumers
Patches to make tests run on older versions or other browsers will be accepted provided they are relatively simple and do not add undue complexity to the test.
Tests should be as cross-platform as reasonably possible, working across different devices, screen resolutions, paper sizes, etc. The assumptions that can be relied on are documented here; tests that rely on anything else should be manual tests that document their assumptions.
Fonts cannot be relied on to be either installed or to have specific
metrics. As such, in most cases when a known font is needed, Ahem
should be used and loaded as a web font. In other cases,
should be used.
Tests must not depend on external network resources. When these tests are run on CI systems, they are typically configured with access to external resources disabled, so tests that try to access them will fail. Where tests want to use multiple hosts, this is possible through a known set of subdomains and the text substitution features of wptserve.
Tests should make it obvious when they pass and when they fail. It shouldn’t be necessary to consult the specification to figure out whether a test has passed of failed.
A number of style rules should be applied to the test file. These are not uniformly enforced throughout the existing tests, but will be for new tests. Any of these rules may be broken if the test demands it:
No trailing whitespace
Use spaces rather than tabs for indentation
Use UNIX-style line endings (i.e. no CR characters at EOL)
We have a lint tool for catching these and other common mistakes. You
can run it manually by starting the
wpt executable from the root of
your local web-platform-tests working directory, and invoking the
lint subcommand, like this:
The lint tool is also run automatically for every submitted pull request, and reviewers will not merge branches with tests that have lint errors, so you must fix any errors the lint tool reports. For details on doing that, see the lint-tool documentation.
But in the unusual case of error reports for things essential to a certain
test or that for other exceptional reasons shouldn’t prevent a merge of a
test, update and commit the
lint.ignore file in the web-platform-tests
root directory to suppress the error reports. For details on doing that,
see the lint-tool documentation.
In order to be included in an official specification test suite, tests for CSS have some additional requirements for: