[JDEV] Namespaces

[David Waite]

Kriggs wrote:

>If you would like to help me (by reading these 90 lines of text), try your 
>best to correct me where I am wrong, and answer my questions (Lines marked 
>with an asterisk (*) are points which I especially am having trouble with):
>
>1)
>
>    <a xmlns="x:y">
>      <b c="value1"/>
>    </a>
>
>In this example, <a/>, <b/> and c="" are all using the x:y namespace. <b/>'s 
>namespace is "x:y", just like <a/>'s, as opposed to my previous theory that 
><b/>'s namespace becomes x:y:a.
>
Yep. Well, there is some debate over whether c="" is in a namespace at 
all. the names recommendation does not clarify on this point until the 
appendix, which is non-normative. Long discussion.

>* Would putting an x:y: prefix to every tag element's name do the same thing 
>as the use of xmlns="x:y" in this example?
>
Incorrect.
1) You must use namespace prefixes to qualify an element not in the 
default (non-prefixed) namespace. You cannot substitute a namespace URI 
for said prefix. There isn't a standard for writing out "this element in 
this namespace".

2) Prefixes cannot contain the ':' character anyways. (see 
http://www.w3.org/TR/1999/REC-xml-names-19990114/#ns-decl)

>2)
>
>    <a xmlns="http://..." xmlns:s="x:y">
>      <s:b c="value1">
>        <d/>
>        <e s:f="value2"/>
>      </s:b>
>    </a>
>
>In this example,
><a/> is using the "http://..." website namespace
><b/> is using the x:y bulltin namespace because of the s: prefix
>c="" is using the x:y bulltin namespace because it is <b/>'s attribute.
><d/> is using the "http://..." website namespace, because it's prefix doesn't 
>depend on it's parent's prefix.
><e/> is using the "http://..." website namespace for the same reason
>f="" is using the x:y bulltin namespace. Although normally an attribute 
>inherits from it's parent, f="" has an s: prefix.
>
Other than the comment above on c="", sure.

>3)
>
>    <a xmlns="v:w">
>      <b xmlns="x:y">
>        <s:c/>
>      </b>
>    </a>
>
><a/> has v:w as it's namespace
><b/> has x:y as it's namespace
>* <c/> has no namespace handler (?)
>
Yep.

>4)
>
>In a more explicit example:
>
>    <stream:stream 
>    to="denmark" 
>    xmlns="jabber:client" 
>    xmlns:stream="http://etherx.jabber.org/streams">
>    [...]
>    <iq type="set" id="1">
>      <query xmlns="jabber:iq:auth">
>        <username>nick</username>
>        <password>pass</password>
>        <resource>home</resource>
>      </query>
>    </iq>
>    [...]
>    </stream:stream>
>
><stream:stream/> uses the xmlns:stream namespace (because of the stream:... 
>part) looking for the element definition of "stream" (because of the 
>...stream part)
>
You mean "http://etherx.jabber.org/streams" namespace.

><iq/>, having no prefix, defaults to the jabber:client namespace.
><query/>, <username/>, <password/> and <resource/> all take from 
>jabber:iq:auth.
>  
>
Yep.

>* Do I take it that jabber, despite the xml standard, applies xmlns:* to the 
>level and lower, but applies xmlns only to the lower levels? Or does every 
>jabber:iq:* contain <query/> in it's definition?
>
There was some confusion over the namespace spec. "query" as above is 
not in the jabber:client namespace, even though it has been reported in 
some documentation as thus before.

>* I notice that jabber:iq:auth has more than 1 colon (:). Is this only 
>possible for namespaces, or can tag elements have that too (namespace 
>declared like xmlns:a:b:c="name:space")? If so, how do you separate the 
>element from the namespace? Is an actual element name not allowed to have any 
>colons (:) in them?
>
Namespaces can have more than one colon, as they are URIs. A qualified 
element or attribute name can only be qualified indirectly by a prefix, 
not directly by a namespace.

-David Waite