RE: [xml-dev] My report on experiments with unused namespaces

["Costello, Roger L." <costello@mitre.org>]

Hi Folks,

I wish to verify that I correctly understand Amelia's post.

EXAMPLE #1

Consider this XML document:

---------------------------------------------------------
<Root xmlns:soap="http://www.w3.org/2001/09/soap-envelope";> 

    <faultcode>soap:Client</faultcode>

</Root>
---------------------------------------------------------

Notice the value of <faultcode> is a QName.

If the application that receives the XML document is _not_ "SOAP aware" then the application _may_ treat soap:Client as simply data, not as a QName. That application _may_ consider this to be an unused namespace:

    http://www.w3.org/2001/09/soap-envelope 

If the application that receives the XML document _is_ "SOAP aware" then the application _must_ process soap:Client as a QName. Thus, that application _must_ consider this to be a used namespace:

    http://www.w3.org/2001/09/soap-envelope

Is this correct?


EXAMPLE #2

Consider this XML document:

---------------------------------------------------------
<soap:Envelope xmlns:soap="http://www.w3.org/2001/09/soap-envelope";> 

    ...

<soap:Envelope> 
---------------------------------------------------------

Notice that the element name is a QName.

According to the XML REC, element names may contain colons. Thus, soap:Envelope could simply be the name of the element. However, the REC also says this:

    Note:

    The Namespaces in XML Recommendation [XML Names] 
    assigns a meaning to names containing colon characters. 
    Therefore, authors should not use the colon in XML names 
    except for namespace purposes, but XML processors must 
    accept the colon as a name character.

I conclude from that note that all XML-aware applications which receive the above XML document _must_ treat soap:Envelope as a QName. Therefore all applications (SOAP-aware applications and SOAP-unaware applications) _must_ consider this namespace as being used: 

    http://www.w3.org/2001/09/soap-envelope 

Is this correct?

/Roger

-----Original Message-----
From: Amelia A Lewis [mailto:amyzing@talsever.com] 
Sent: Wednesday, September 15, 2010 7:38 PM
To: Costello, Roger L.
Cc: xml-dev@lists.xml.org
Subject: RE: [xml-dev] My report on experiments with unused namespaces

On Wed, 15 Sep 2010 17:45:10 -0400, Costello, Roger L. wrote:
> [Definition] Used Namespace: a namespace in an XML instance document 
> which is 
> 
> (1) used in an element name, or 
> 
> (2) used in an attribute name, or 
> 
> (3) used in a QName of an attribute value, or 
> 
> (4) used in a QName of an element value.

You have a problem.

While the combination:

xmlns:a="uri" in an element followed by "a:something" as attribute or 
element content *suggests* that the namespace bound to the prefix 'a' 
is in use--indeed, it is unusual to encounter letter-colon-letter 
without whitespace, and this might be regarded as indicative of 
namespace-prefix reference--it is by no means a guarantee.  In order to 
be certain, you have to know more than "this is XML"; you have to know 
"this is an XProc attribute representing magic-word," or "this is an 
XSLT attribute representing selection" or "this a a GUE element 
representing that the referenced namespace has been eaten by a grue".

In short: because XML namespaces were, very early, used in element and 
attribute content, but you cannot tell whether element or attribute 
content contains references to XML namespaces in scope, you can't tell 
what's going on unless your processor actually processes this dialect 
of XML.  This is one of the most depressing problems in XML.

What's worse: text-extraction, or naive XML object-model extraction, 
can create a document in which:

<grue>
  <dark>gue:eaten</dark>
</grue>

is meaningless.

(The above is potentially extracted from:

<map xmlns="infocom://great.underground.empire/" 
xmlns:gue="infocom://great.underground.empire/">
  ...
  <grue>
    <dark>gue:eaten</dark>
  </grue>
</map>

Note that this also raises a potentially significant issue in terms of 
namespace minimization: the namespace-in-content above is using a 
prefix (because it *must* use a prefix, for XPath, for instance; you 
cannot bind the default prefix in an XPath 1.0 expression), but it is 
otherwise redundant.  The authors want to use the default prefix 
*except* where forced to use the non-default prefix.  So they have 
bound the same namespace to two different prefixes (one of them nearly 
invisible).

Amy!
-- 
Amelia A. Lewis                    amyzing {at} talsever.com
Love doesn't just sit there, like a stone, it has to made, like bread,
remade all the time, made new.
                -- Ursula K. Le Guin