W3C home > Mailing lists > Public > public-webapi@w3.org > March 2006

Re: ACTION-87: Selectors API

From: Ian Hickson <ian@hixie.ch>
Date: Wed, 22 Mar 2006 22:48:51 +0000 (UTC)
To: Anne van Kesteren <annevk@opera.com>
Cc: "Web APIs WG (public)" <public-webapi@w3.org>
Message-ID: <Pine.LNX.4.62.0603222228210.315@dhalsim.dreamhost.com>

On Wed, 22 Mar 2006, Anne van Kesteren wrote:
>
> http://dev.w3.org/cvsweb/~checkout~/2006/webapi/selectors-api/draft/selectors-api.htm

New comments:

The term "first" is not defined (for |match()|). Specifically, it should 
say whether it is the first match in a depth-first pre-order traversal, or 
whether it is a breath-first traversal (depth-first pre-order makes the 
most sense, IMHO).

"In document order" could be more explicit.

There is no normative description of how |Function| objects in ECMAScript 
end up supporting XPathNSResolver.

I don't think "In ECMAScript bindings the nsresolver argument in both 
match and matchAll must be an optional argument." as a requirement makes 
sense. I think it would be better to phrase it as something like "In 
ECMAScript, if the nsresolver argument in an invocation of match() or 
matchAll() is omitted, UAs must handle the invocation as if the nsresolver 
argument was null." or something. That text sucks but the point is that 
the requirement is on the UA, and the requirement clearly states what 
must happen. As currently phrased, the spec doesn't say what it means for 
the argument to be optional.

Also, the spec should say what should happen if the nsresolver argument is 
null.


> > * Having an interface doesn't imply behaviour -- e.g. NodeList doesn't 
> > imply that NodeList is live. You can have an object that implemnets 
> > NodeList and is not live.
> 
> DOM Level 3 Core says it's live. Per discussion on public #webapi I see 
> that is suboptimal and if DOM Level 3 Core gets errata to make that more 
> clear I'll reconsider it.

If you really want to not use the term NodeList, I recommend defining 
StaticNodeList as:

   typedef StaticNodeList NodeList;

...rather than duplicating the interface definition.


> > * I would recommend against supporting namespaces in the first 
> > version, for simplicity.
> 
> I agree that they are probably not that much needed by authors. 
> Therefore the argument is optional in ECMAScript bindings so it doesn't 
> harm the scenario were people probably use it the most.

The API supports namespaces. That makes it more complicated, whether or 
not the author can omit the namespace argument.


> > * The spec doesn't say _how_ to resolve the namespaces using the 
> > nsresolver argument.
> 
> I deferred this to DOM Level 3 XPath now.

That is a non-normative note. It's also not clear where in that spec it is 
defined what the requirements on implementations and authors are.


> > * IMHO the argument to getElementsBySelector should be a "group of 
> > selectors" not a "selector" (using Selectors terminology).
> 
> Used this terminology although I don't really see the Selectors draft 
> defining this term (using <dfn> or whatever). Could you please point it 
> out?

Section 5 of Selectors.


> > * IMHO the method should not raise an exception when the selector 
> > contains a pseudo-element. It should would return an empty list.
> 
> Given that it per definition only returns Element nodes I don't see why 
> it shouldn't raise an exception.

There are three main reasons. One is that pseudo-elements can in certain 
cases refer to Elements (XBL2 in particular introduces this concept).

The second reason is that pseudo-elements are perfectly valid selectors, 
so if the script is just running through a list of selectors and checking 
each one for a match, it will not want valid selectors to throw an 
exception. Consider, e.g., if the input is coming from a user -- you want 
to distinguish "doesn't match any elements (because it can't)" from "is 
invalid syntax".

The selector "foo:not(foo)" will never match anything either, but it 
doesn't raise an exception. You don't want to require the implementation 
to look at the selector any more than it would for a normal CSS 
implementation.

The third reason is that you could have a selector that contains 
pseudo-elements and yet can match an element, for instance:

   p, p::first-line

...can match an element (p).


Otherwise, looks good.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'
Received on Wednesday, 22 March 2006 22:48:59 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 14:18:54 GMT