[css-pseudo-4] UA must (instead of should) use the OS-default highlight colors ...

[TalbotG]

If "Historically (and at the time of writing) only color and background-color have been interoperably supported."
coming from §3.2 Styling Highlights
https://www.w3.org/TR/css-pseudo-4/#highlight-styling

then I think the following statement can safely use a MUST and not a SHOULD:

"
The UA should use the OS-default highlight colors for ::selection when neither color nor background-color has been specified by the author.
"
coming from §3.4 Cascading and Per-Element Highlight Styles
https://www.w3.org/TR/css-pseudo-4/#highlight-cascade

[frivoal]

OS-default highlight colors are not well defined, not really testable, and not even necessarily a thing that exists on all platforms. General purpose OSes will certainly have something, but special purpose devices might not.

All in all, I think “should” is fine, but I'll also note that this text is likely to evolve as a consequence of #4720

[TalbotG]

OS-default highlight colors are not well defined,

OS-default highlight colors can be user-settable, therefore they can not be predictable. This was the case in Windows XP and it is the case in KDE3 and KDE4 (and probably other desktop environments and possibly with browser themes) in Linux.

not really testable

The repository has now 2 tests:
http://test.csswg.org/suites/css21_dev/nightly-unstable/html4/system-colors-001.htm
and
http://wpt.live/css/CSS2/ui/system-colors-001.xht
which is a manual test
and
http://wpt.live/css/css-color/parsing/system-color-valid.html

, and not even necessarily a thing that exists on all platforms. General purpose OSes will certainly have something, but special purpose devices might not.

I do not know about other OSes, other platforms and other devices. So I can not comment on this.

All in all, I think “should” is fine, but I'll also note that this text is likely to evolve as a consequence of #4720

My understanding is that #4720 only relates to presence of fill/stroke ... which only apply to SVG objects.

[fantasai]

@TalbotG is correct, the wording here is not as tight as it should be. I've switched it out to:

The UA must use its own highlight colors for ''::selection'' only when (and only when) neither 'color' nor 'background-color' has been specified by the author.

I believe this is what was meant.

There's an additional expectation that these colors should match the OS colors, but I'm not sure we need to impose that as a requirement?