Follow-up: Public Comment #191 on Namespaces vs. Prefixes

["Dennis E. Hamilton" <dennis.hamilton@acm.org>]

On Public Comment #191, I have the action to see if this comment has been addressed in ODF 1.2 (cd01).  The comment concerns the normative versus exemplary status of the namespace prefixes used throughout the specification.

Based on my analysis, this situation has not been resolved.  I recommend resolving this comment in ODF 1.2 by changes for the next Committee Draft.

I accept an action item to propose revised text along with a policy to be applied to mentions of namespaces, qualified names (syntactically: QNames), namespace prefixes, and local names throughout the ODF specification.  (I also recall a suggested phraseology in a note from Robert Weir somewhere on a list.)

 - Dennis

[1] Dennis E. Hamilton: OpenDocument 1.2v7-02: 1.3 Namespaces.  Submitted to office-comments@lists.oasis-open.org on 2008-05-17.  Available at <http://lists.oasis-open.org/archives/office-comment/200805/msg00029.html>.

[2] Patrick Durusau: Reply to [1] via office-comments@lists.oasis-open.org on 2008-05-18.  Available at <http://lists.oasis-open.org/archives/office-comment/200805/msg00035.html>. 


ANALYSIS
--------

1. The ODF 1.3 Namespaces section provides four tables, each having columns for Prefix, Description, and Namespace.

The only text consists of the following statement and accompanying note:

    The namespaces used or defined by OpenDocument are 
    listed in tables 1-4.

    Note: XML namespaces are defined by the ''Namespaces 
    in XML'' specification [xml-names].

2. In ODF 1.1, the same table format is used but the opening paragraph makes unfortunate reference to "their default prefixes" for the namespaces.  I.e.,

   Table 1 lists the namespaces that are defined by the
   OpenDocument format and their default prefixes. For 
   more information about XML namespaces, please refer 
   to the Namespaces in XML specification [xml-names].

3. DIFFERENCES

The removal of any mention of default namespace here is valuable.  

However, there is now no indication of the significance of the Prefix columns of the tables.  

Neither statement referencing [xml-names] is very satisfactory considering the normative force of [xml-names] with regard to XML document formats established in the specification.

4. ADDITIONAL PROBLEMS

In his reply to [1], Patrick Durusau expresses a number of related concerns around the prefixes being taken as normative by implementations and around the confusion of namespaces and their prefixes elsewhere in the specification [2].  

I believe a proper course for clarification of this aspect of the specification would be as follows:

4.1 Assertion that [xml-names] applies to all XML documents defined in those schemas introduced as part of the OpenDocument specification and those documents shall conform to [xml-names].

4.2 Informative note that the only conditions on the choice of prefixes and presence of namespace declarations are those established in [xml-names].  The prefixes identified in the tables and employed throughout the specification, in examples, and in the schemas, have been chosen for consistency of exposition and are non-normative beyond identifying the applicable namespace in those occurrences.  (One could go so far as to mention the xmlns declarations that those cases are assumed to be in the scope of.)

4.3 For ODF 1.2 it is probably worth mentioning in 1.3 that namespace prefixes are used in two additional ways beyond those established in [xml-names].  One special case is prefixes from namespace declarations being used at the head of attribute values to signify a custom [i.e., foreign] value.  The other case is the use of prefixes from namespace declarations as prefixes in attribute values understood as having RDF CURIE shorthand for URIs based on those namespaces.

4.4 It might also be worth noting that, in accordance with [xml-names] unprefixed attribute names are not local names of any namespace.  Also, in accordance with [XML 1.0], 

4.5 Patrick identified a misleading statement in the metadata section (from ODF 1.1 section 3):
 
    "The metadata elements borrow heavily upon
     the metadata standards developed by the 
     Dublin Core Metadata Initiative 
     (http://www.dublincore.org).  Metadata
     elements drawn directly from the Dublin
     Core work use its namespace prefix (see
     section 1.3).

The phrasing in ODF 1.2 Part 1 cd01 section 3.3.1 (3rd paragraph) is

     The pre-defined metadata elements borrow
     heavily upon the metadata standards 
     developed by the Dublin Core Metadata 
     Initiative (http://www.dublincore.org). 
     Metadata elements drawn directly from the
     Dublin Core work use its namespace prefix. 
     See 1.3.

The word prefix should be stricken.  An easy way out is to strike the entire statement (also eliminating the in-line URL), since the section 1.3 table says what is necessary and the schema and the individual descriptions already (under-) specify the reliance on specific DCMI elements. In addition, there are other places where the Dublin Core elements occurences are provided in the ODF Schema, completely apart from the arbitrary use available in RDF metadata URIs for properties.

4.5 Patrick identifies other places where there is casual wording around namespace and prefix agreement.  These should be reviewed and a consistent policy applied.  The main solution seems to be to not say so much.  



 - Dennis

Dennis E. Hamilton
------------------
NuovoDoc: Design for Document System Interoperability 
mailto:Dennis.Hamilton@acm.org | gsm:+1-206.779.9430 
http://NuovoDoc.com http://ODMA.info/dev/ http://nfoWorks.org