W3C home > Mailing lists > Public > public-css-archive@w3.org > December 2017

Re: [csswg-drafts] [selectors] :link and :visited are not mutually exclusive in implementations

From: François REMY via GitHub <sysbot+gh@w3.org>
Date: Thu, 07 Dec 2017 04:27:20 +0000
To: public-css-archive@w3.org
Message-ID: <issue_comment.created-349859523-1512620840-sysbot+gh@w3.org>
@victoriasu So, I just want to take an opportunity to clarify my thinking.

I do believe that if we keep the spec text as it is today, we should at the very least add a note mentioning that for most properties, :link will apply regardless of visited status, and point to [DBaron's spec](https://dbaron.org/mozilla/visited-privacy). I understand @dbaron's point of view that says that, for the purpose of the selector matching, :link and :visited are mutually exclusive (as in you cannot ever have :link:visited match), I just think that for an author reading the spec, it is important to understand that :link will, in practice, match all links except for a few select properties.



To return to the interop issue highlighted in the first post of this thread:

Globally, how it works in Edge is we kinda have a "color" and a "visited-color" property, and when a style rule applies "color:red" we set both of them; if the style came from a :visited rule however, we only set "visited-color" (we do this for a limited subset of properties, for other properties, a visited rule will not set anything when applied; converserly, when a rule contains :link we should only set the non-visited property, though whether we do this for all properties is unclear). This means that :link applies all the times for most properties, and does not apply for properties for which we got the ":link sets only non-visited field" rule right (when we do not :visited rules are still allowed to override these :link declarations that also happen to have set the visited-property, they just need higher specificity).

It looks like `background-color` is a property for which we didn't get the ":link does not set the :visited part" right, hence the interop difference in the demo above. For all other properties I tested, we behave like other browsers, so it is just a bug on us to fix this issue.

To conclude, I think as a result of this, only an editorial change should be necessary: adding a note to the spec would be great, if only because the current text caused enough confusion for browser engineers to file this issue in the first place.


-- 
GitHub Notification of comment by FremyCompany
Please view or discuss this issue at https://github.com/w3c/csswg-drafts/issues/2037#issuecomment-349859523 using your GitHub account
Received on Thursday, 7 December 2017 04:27:23 UTC

This archive was generated by hypermail 2.4.0 : Tuesday, 5 July 2022 06:41:21 UTC