W3C home > Mailing lists > Public > www-style@w3.org > February 2015

Re: element existence selector

From: Tab Atkins Jr. <jackalmage@gmail.com>
Date: Fri, 27 Feb 2015 08:40:17 -0800
Message-ID: <CAAWBYDAVUgUhajoCaW3ZA+=POP5ia2=1CCk=9TomccUFgS9iAw@mail.gmail.com>
To: Peter Krauss <ppkrauss@gmail.com>
Cc: www-style list <www-style@w3.org>
On Fri, Feb 27, 2015 at 6:37 AM, Peter Krauss <ppkrauss@gmail.com> wrote:
> Since CSS2 we can express the "existence of attributes",
>    http://www.w3.org/TR/WD-CSS2/selector.html#h-5.7.1
> with the [att] selector,
>     element[att] {do}
> and it is easy implement this kind of conditional selector in CSS parsers...
> The suggestion (or question) here is about use the same ideia with elements,
> a kind of existence operator,
>     e1 <existenceOperator> e2 {do}
> to say "if element e1 exists, selects the e2 element".
> * if it is a new suggestion, can I discuss here this subject?
> * if there are some old discussion about it, sorry, can somebody reply with
> a link?
> ------
> PS: I think that this kind of <existenceOperator> can used in a big subset
> of (imagined) applications of the problematic previous-sibling-selector, and
> also used as a "trigger" in event selectors, like
>       e1:hover <existenceOperator> e2 {do}

Ah, now I get it.  You're splitting the selector into two completely
separate parts; if the first part matches an element, you run the
second part to actually select something.  More like "if(e1) then e2".

So, in the full-powered :has() world (which will only exist for
querySelector and similar APIs) this is just `:root:has(e1) e2`
(assuming that no part of e2 is supposed to match the root; otherwise
you have to merge the first compound selector of e2 into the :root

Unfortunately, this won't get into the fast profile (what's supported
in stylesheets) for the same reason that full-powered :has() won't;
it's way too slow, and makes invalidating styles too difficult, since
a change *literally anywhere in the document* can affect whether the
selector matches the e2 element.

The specific use-case you pointed to in your link might be possible,
though.  We agreed to add some pseudo-classes to help select cells in
the same column of a table; selecting cells in the same row is usually
handled by the sibling combinators, but the presence of rowspan does
mess things up a bit.

Received on Friday, 27 February 2015 16:41:11 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:52:01 UTC