-webkit-border-radius isn't alias in WebKit

[gsnedders]

Differs from border-radius only in the interpretation of a value consisting of two lengths: '-webkit-border-radius: l1 l2;' is equivalent to 'border-radius: l1 / l2;'.

This implies one, three, and four length values are identical? Needs investigation, though.

[Loirooriol]

It's not a pure alias in Blink either:

var {style} = document.createElement("div");
style.cssText = "-webkit-border-radius: 1px 2px";
style.cssText; // 'border-radius: 1px / 2px;'

This implies one, three, and four length values are identical?

Yes, the grammar of -webkit-border-radius is identical to border-radius except when exactly two space-separated values are provided. See parser code for Blink and WebKit.

BTW, this implies that serializing -webkit-border-radius for border-radius: 1px 2px should not be 1px 2px in Blink/WebKit. Most reasonable candidates are 1px 2px 1px, 1px 2px 1px 2px and 1px 2px / 1px 2px, probably it should be the 1st one which is the shortest one. But Blink uses the wrong 1px 2px, and WebKit never serializes -webkit-border-radius.

[karlcow]

So results for this code (yes with -webkit- prefix):

var {style} = document.createElement("div");
style.cssText = "-webkit-border-radius: 1px 2px";
style.cssText;

• WebKit: border-radius: 1px 2px;
• Gecko: border-radius: 1px 2px;
• Blink: border-radius: 1px / 2px;

[Loirooriol]

The serialization for border-radius was completely broken in WebKit, unrelated to -webkit-border-radius.
I just fixed it, now it produces 1px / 2px.
Gecko says 1px 2px since it treats -webkit-border-alias as a pure alias.

[karlcow]

To link dots in this discussion (paper trail and context)
WebKit/WebKit@fd9594c
WebKit/WebKit#6820
https://bugs.webkit.org/show_bug.cgi?id=183994