RE: [cgmo-webcgm] Namespace declaration ISSUE

[Lofton Henderson <lofton@rockynet.com>]

At 09:39 AM 8/22/2005 -0700, Cruikshank, David W wrote:
>Here's what S1000D says about declaring namespace in 7.4.1.1.  Later in 
>the spec it also says that all namespaces declared must be registered.

This little PDF snippet is good -- recommended reading.  It is a brief and 
lucid description that confirms what it took me lotsa' hassling to sweat 
out of "XML Namespaces".

Where does one register namespaces?  Is there some sort of registry, like 
the MIME type registry?  Or is this an ASD thing?

>I also checked with one of our namespace "experts".  His preference is 
>REQUIRED.

Why?  Is it only a matter of preference, or does he have some substantive 
pros/cons for choosing #REQUIRED?  Benoit chose the other option (#FIXED), 
with an interesting observation about a disadvantage of the #REQUIRED option...

At 03:04 PM 8/22/2005 -0400, Benoit Bezaire wrote:
>[...]
>b) Opt 2. Why? Because...
>  - I think we should follow what other standards have done.
>  - I don't think the limitation you point out will ever be used.
>  - Opt 1 has no way of making sure that xmlns="" points to the WebCGM
>  namespace. Example, opt 1 actually makes this a valid file:
>
>  <webcgm xmlns="notawebcgmnamespace">
>    <grobject.../>
>  </webcgm>
>
>  which is wrong.

-Lofton.

>-----Original Message-----
>From: Lofton Henderson [mailto:lofton@rockynet.com]
>Sent: Saturday, August 20, 2005 2:13 PM
>To: cgmo-webcgm@lists.oasis-open.org
>Subject: [cgmo-webcgm] Namespace declaration ISSUE
>
>
>I would like this on the agenda of the next telecon.  And would
>namespace-knowledgeable people please comment in advance?
>
>[1] http://lists.oasis-open.org/archives/cgmo-webcgm/200508/msg00010.html
>
>Basically, as I explained in [1], I think what we had in the spec preceding
>the current CD2 text was broken.  While it required definition of
>application namespaces, it also allowed the WebCGM namespace to be
>undefined, which I believe violates "XML Namespaces".
>
>So I tossed in something to make it legal in the CD2 text -- every XCF
>instance MUST define the WebCGM namespace (which makes all of our test
>cases illegal).
>
>As I explained in [1], this is one of two options I see for a
>solution.  The other would effectively default the WebCGM namespace, but
>with a side effect that I describe in [1] (this solution apparently matches
>what SVG1.1 does).
>
>So ...
>
>a.) Is my analysis correct in [1]?
>b.) Which option do you prefer?
>
>-Lofton.
>