Re: [selectors] feedback

On 08/27/2015 10:21 AM, Anne van Kesteren wrote:
> On Thu, Aug 27, 2015 at 9:41 AM, fantasai <> wrote:
>> Right, this text doesn't appear in the section on :empty.
> I don't understand how that makes sense.

... I don't understand what's confusing. The things that skip
non-element nodes specify that they skip non-element nodes.
Things that care about non-element children say what they care
about. The model of what's significant might be inconsistent,
but everything's specified explicitly.

>> Perhaps you're misunderstanding what I meant. DOM provides the entire
>> DOM tree for pattern-matching--of course. But for its .query methods
>> the DOM spec compiles the list of elements that are potential output
>> and filters them by whether or not they match. Selectors only defines
>> what it means for an element to match.
> How does that solve anything for pseudo-elements, which are not
> elements and cannot be matched?

The algorithm currently specified in "evaluate a selector", which
you are invoking in DOM, returns pseudo-elements in addition to
elements unless otherwise specified. And DOM does not otherwise

(Things like this is why I'm advocating for just using "match" as
the spec interface.)

>> This way Selectors doesn't need to decide what elements are valid
>> output, or define how to traverse the DOM tree to find said potential
>> output elements, or define out the output of .querySelector is sorted
>> or otherwise represented. These are particulars of the .query methods,
>> not particulars of how selectors work, so they belong in DOM, not in
>> Selectors.
> Yeah maybe. It's somewhat annoying that we're revisiting this setup a
> good year after we decided on something. Feel free to provide PRs
> here:

Okay. Will argue this with dbaron and Tab and see if we can settle on
some wording.

(Sorry about the late review. I didn't have time to keep up with
Selectors year ago.)


Received on Thursday, 27 August 2015 11:09:00 UTC