260337 2023-08-17 01:57:47 -0700 new URL("file:///path").origin returns "file://" instead of "null" 2023-08-18 04:02:05 -0700 1 1 1 Unclassified WebKit New Bugs Safari 17 Unspecified Unspecified NEW BrowserCompat, InRadar, WPTImpact P2 Normal --- 128023 1 karlcow webkit-unassigned annevk webkit-bug-importer oldest_to_newest 1971770 0 karlcow 2023-08-17 01:57:47 -0700 Some of the fails in these series of tests. https://wpt.fyi/results/url?label=master&label=experimental&aligned&q=safari%3Afail%20chrome%3Afail%20firefox%3Apass In Safari and Chrome new URL("blob:file://host/path").origin "file://" new URL("file://host/path").origin "file://" new URL("file:///path").origin "file://" in Firefox new URL("blob:file://host/path").origin "null" new URL("file://host/path").origin "null" new URL("file:///path").origin "null" Example https://wpt.fyi/results/url/a-element-origin.html FAIL message: assert_equals: origin expected "null" but got "file://" Cf https://url.spec.whatwg.org/#special-scheme https://url.spec.whatwg.org/#concept-url-scheme https://url.spec.whatwg.org/#origin The specs says for the file scheme > Unfortunate as it is, this is left as an exercise to the reader. > When in doubt, return a new opaque origin. which seems to imply to return null. but this sentence is probably not necessary and confusing: " Unfortunate as it is, this is left as an exercise to the reader. " 1971771 1 webkit-bug-importer 2023-08-17 01:57:54 -0700 <rdar://problem/114015467> 1972010 2 annevk 2023-08-18 03:03:55 -0700 I'm changing the tests in https://github.com/web-platform-tests/wpt/pull/41158 which would resolve the WPTImpact. I also don't think there's BrowserCompat here as at least Chromium does the same thing. We might still want to fix this though. Returning "file://" is not exactly helpful. Might have to be done at the point where we actually serialize the origin though to not upset origin comparisons and such.