Re: [selectors-nonelement] ::attr(*|localname), ::attr(ns|*), and ::attr(*)

On Wed, 26 Feb 2014 14:37:22 +0100, Jirka Kosek <jirka@kosek.cz> wrote:

> On 26.2.2014 14:30, Simon Pieters wrote:
>> Does SVG or DocBook say that an attribute in the XLink namespace is
>> translatable by default?
>
> No, only HTML5 defines application specific rules. For all other
> languages default is simple -- attributes are not translateable, and
> elements are translateable. But please note that ITS is not only about
> translateablity,

Is there a non-translation ITS use case for ::attr(*) involving namespaced  
attributes?

> and that non-element selectors should extend selectors
> to be more generic selection/query language, not only to cater for some
> ITS use-cases.

I don't see how this argues for one behavior over the other.

To sum up what we have so far:

* The stated use cases so far would work equally well with either behavior.
* Jirka thinks selecting all attributes better matches author expectations  
or is more useful.
* Selecting all attributes is inconsistent with ::attr(foo) and [foo]
* Selecting all attributes is inconsistent with the universal selector  
when a default namespace is declared.

So, unless there are further use cases that would tip the scales, it seems  
it boils down to what is considered more important: consistency or  
matching author expectations. Or maybe research/poll what authors expect,  
in case their expectation is different from Jirka's?

-- 
Simon Pieters
Opera Software

Received on Wednesday, 26 February 2014 14:17:07 UTC