W3C home > Mailing lists > Public > www-dom@w3.org > October to December 2011

Re: setAttributeNS(): should it change the namespace prefix?

From: David Flanagan <dflanagan@mozilla.com>
Date: Tue, 29 Nov 2011 10:17:26 -0800
Message-ID: <4ED521B6.8090302@mozilla.com>
To: www-dom@w3.org
On 11/29/11 8:22 AM, Anne van Kesteren wrote:
> document.head.setAttributeNS("test", "x:a", "a")
> document.head.setAttributeNS("test", "e:a", "b")
> Should the result be an attribute "x:a" or "e:a"? Gecko says "e:a", 
> Opera/Webkit say "x:a". Making the prefix member of an attribute 
> completely immutable seems somewhat preferable to me. Opinions?
Step 10 of the setAttributeNS() spec currently requires e:a, right?

That's how I implemented it in dom.js, and it made my attribute code and 
data structures more complicated than they would have been otherwise.  
If I'm not mistaken, it is that one step in this one method that allows 
an element to have multiple attributes with the same qualified name. So 
this one method causes the specs for all the other attribute methods to 
have to say "...the first attribute in the context object's attributes 
whose name is..." instead of "...the attribute whose name is..."

Making the change would simplify the spec and simplify implementations 
and presumably (given Opera and Webkit) wouldn't have much web 
compatibility impact, so I'm for it.  (Though I wish I hadn't wasted my 
time getting it right in the first place!)

But even if we can't change the spec for setAttributeNS, can we 
deprecate it and the rest of the *AttributeNS methods and replace them 
with something simpler as part of the "improving the DOM" effort?  For 
example, we could allow setAttribute() to be called with two arguments 
(qname, value) or three arguments (ns, lname, value).  And similarly for 
getAttribute, hasAttribute and removeAttribute.


Received on Tuesday, 29 November 2011 18:18:04 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 20 October 2015 10:46:18 UTC