I was unable to find any references stating that WebKit's behavior at bugs.webkit.org/show_bug.cgi?i… violates any such spec

I'll still have to look into the remaining ones, but I am curious if you have any references that state that it's a spec violation (I skimmed the W3C IRC logs, the MQ5 spec, the W3C GitHub issue, etc)

visit nouveau.freedesktop.org, open the developer tools, go to the "Sources" tab, open "style.css", remove the line containing "--logo-path: url('logo-light.svg');" (line 55), and replace the line from "--logo-path: url('logo-dark.svg');" with "--logo-path: url('logo.svg');" (line 17)

Then, switch between light and dark styles, and you'll notice that the SVG's text doesn't adapt

("logo.svg" is the SVG file that embeds prefers-color-scheme to adapt to light and dark styles)

nouveau · freedesktop.org

^{freedesktop.org}

@ity to me, not supporting a standard is the equivalent of not aligning with it. So to answer your question, it's an example of WebKit failing to align with web standards

That said, since "align" seems to be ambiguous, it could as well be "failed to align with or support established web standards"

I cannot find it

I can only find a link to the WebKit bug tracker, then two SVG links, plus ofc a link to your blog post, which I read and I cannot find any links to any standards there either

The WebKit bug tracker links to a W3C discussion page with inconclusive results (as far as I can tell, they arrived at the conclusion that SVGs MAY inherit the value, depending on per-browser decisions). I cannot find anywhere in the Media Query Level 5 specification, which defines the semantics of the prefers-color-scheme media query, a statement that it MUST inherit its value. Just what its value means without further specification of its behavior, plus one big red box stating to be careful with implementing it due to fingerprinting considerations.

(Highlighted keywords MAY and MUST in this fedi post to be interpreted as-defined in RFC2119 - the spec & discussions do not seem to use RFC2119 in parts relevant to the prefers-color-scheme media query)

So, I am truly at a loss here. The prefers-color-scheme Media Query is

• Implemented
• Behaves how the W3C specification defining it prescribes it should
• Does not behave in a predictable manner
• Does not behave in the same way as Chromium & Firefox do

I might be missing some different specification or a part of the MQ5 specification that states that it SHOULD or MUST behave the way you and many others are expecting, which is why I am asking if you know of any, since you are stating it does not support a specification properly

the only thing I can do is link to developer.mozilla.org/en-US/do…. Other than that, I don't know what else there is to say...

> plus ofc a link to your blog post, which I read and I cannot find any links to any standards there either

Ah, sorry for that. The blog post isn't used as a source/reference to prove a point. It's just a link in case people wanted to read about the rewrite. I removed that link

prefers-color-scheme - CSS | MDN

The prefers-color-scheme CSS media feature is used to detect if a user has requested light or dark color themes. A user indicates their preference through an operating system setting (e.g., light or dark mode) or a user agent setting.

^{MDN Web Docs}

That's not normative, that's descriptive of Firefox & Chromium behavior (the purpose of mdn is to be descriptive rather than normative, unlike the specifications themselves)

MDN even warns that "Respects color-scheme inherited from parent" is "non-standard" and to "Expect poor cross-browser support"

The feature itself is listed as "Full support" for Safari/WebKit on that site.

The feature is also considered "unfinished" by W3C, and W3C in the specification warns that

Information about a user can be used as an active fingerprinting vector. Analysis of impact pending, more information to be provided before spec is published.
User agents and developers implementing this specification need to be aware of this vector and take it into consideration when deciding whether to use the feature. Specifically prefers-reduced-motion, prefers-color-scheme and prefers-reduced-data are currently of concern for exploitation.

W3C further comments on the specific feature that

[css-mediaqueries] Should prefers-color-scheme in SVG images be context-dependent?
RESOLVED: Have prefered-color-scheme reflect 'color-scheme' on the embedding element in the embedding document, to the extent acceptable from security standpoint (pending security review)

There seems to be a draft asking for it in "Secure Animated mode" -> github.com/w3c/csswg-drafts/co…

Which has not made it to the current latest Working Draft of MQ5 (w3.org/TR/mediaqueries-5/), and is only available in the Editor's Draft (drafts.csswg.org/mediaqueries-…)

MQ5 itself is not finalized and is very much experimental.

This has been okayed into being merged into WebKit at 2022 by one of the WebKit maintainers github.com/w3c/csswg-drafts/is…

I have no idea what policy WebKit has for standards, but the only standard it seems to violate is one that begins with this statement:

This is a public copy of the editors’ draft. It is provided for discussion only and may change at any moment. Its publication here does not imply endorsement of its contents by W3C. Don’t cite this document other than as work in progress.

So while it is certainly dumb that it has not been fixed yet, it feels a bit of a stretch to call it a violation of web standards, imo.

I wonder how difficult it is to fix, I've been meaning to try to get into WebKit stuff

[css-mediaqueries] Should prefers-color-scheme in SVG images be context-dependent?

Consider this test-case: for (let img of document.querySelectorAll("img")) { im...

^{emilio (GitHub)}

🥺 always, mew ! I got nerd sniped haha

I even cloned WebKit (it's compiling now) and I'll see if I can patch it. Unsure. Hope it isn't more complex than Firefox.