Created attachment 270350 [details]
which is talking about whether 0 should be handled as 0deg.
However, it's even worse. See the examples in the attached file, where
linear-gradient(95deg, red, blue) is ok
linear-gradient(95, red, blue) is parsed with 3 colours.
smfr says there are also differences in quirks v standards mode
Implementing https://quirks.spec.whatwg.org/#the-hashless-hex-color-quirk per spec would maybe fix this.
cdumez, want to take a stab at this? :-)
Chromium passes the tests
Possibly related: bug 155874
There is no difference in attached test case across all other browsers (Chrome Canary 108, Firefox Nightly 107 and Safari Technology Preview 154).
First two square / rectangular has gradient from top to bottom starting from bad and third square / rectangular from the right side to left with same pattern.
Is something further required? Thanks!
Let's say "no", as identical rendering is interoperable. Someone can file a follow-up bug if needed.