This page shows web-platform-tests project tests which fail in exactly one browser engine. For example, if a test fails in Blink (the engine behind Chrome and other Chromium-based browsers) but passes in Gecko (Firefox) and WebKit (Safari), it is counted here.
Fixing these failures is a very concrete way to improve the overall interoperability of the web for all web developers.
This graph shows the number of engine-specific failures over time, for Blink (Chromium browsers),
Gecko (Firefox), and WebKit (Safari).
The y-axis is the number of tests that fail in exactly one engine, normalized to account for subtests (to learn
more about subtests and how they're normalized here, see the source
code). The x-axis is the date of the test run.
The data for this graph comes from this CSV file which is generated by the web-platform-tests/results-analysis repo.
This list is computed live from the wpt.fyi search API at the
test level (a test counts if exactly one engine fails it).
The historical graph above is subtest-normalized, so the two will not
match exactly. This list is the actionable view of which
tests differ.
The list below is sorted by the number of failures in each directory, so the most important directories to fix are
at the top.