ISSUE: is metadata enumeration an XCF use case?

[Lofton Henderson <lofton@rockynet.com>]

WebCGM TC,

We need to clarify this, so I'm queuing an issue.  Discussion?  Preferences?

ISSUE:  is full metadata enumeration a supported XCF use case?

ALTERNATIVES:

Alt.1 -- YES, full metadata enumeration is a critical use case for WebCGM 2.0.
Alt.2 -- NO, don't try to solve this use case for WebCGM 2.0.

RECOMMENDATION:  Alt.2, NO.

(If NO, should we fix the text by removing the subject use case, or by 
modifying it?  E.g., modify it to indicate that *partial* metadata 
enumeration, e.g., of 'linkuri's and 'content' APS Attrs, would be another 
possible application.  And change it to be example 3-of-3, instead of 
example 1-of-3.)

DISCUSSION:

Dieter's Ch.2 comments started a thread:
http://lists.oasis-open.org/archives/cgmo-webcgm/200506/msg00142.html

At 05:53 PM 6/2/2005 +0200, =?us-ascii?Q?Dieter__Weidenbruck?= wrote:
[...]
>section 2.6  1.)
>Is there a normative section describing this use case?

The wording at section 2.6 #1 is:  "can be used as a scaled down XML 
representation of a WebCGM illustration by enumerating the Application 
Structures IDs, types and attributes."  Those are also exactly the words 
preceding example 4.1 in the normative XCF chapter.

Dieter further pointed out:  if we really  want to use the XCF this way we 
should probably flag the file accordingly to distinguish it from "normal" 
XCFs. We need to think about a way then to add the "name" attribute to such 
a list file. Right now we are not able to enumerate all APS IDs, types, and 
attributes: the name(s) would be missing.

A "normal" XCF is one that is designed to bind metadata to objects in the 
metafile, both standardized metadata like linkuri and private application 
metadata.  If XCF is also to be a metadata enumeration of stuff in the 
file, there are at least two issues that we don't adequately address now:

1.) the 'name' APS Attribute cannot be externalized to the XCF currently.
2.) One probably would not want to loadAndApply such a file.  (It is 
redundant with stuff already in the metafile, or if not totally redundant, 
then it's not really an enumeration.)

"WebCGM 2.0 Requirements" [1] says:
[1] http://www.cgmopen.org/technical/WebCGM_20_Requirements.html#L568

>2.6.1 Capabilities and Usage Scenarios
>
>Below is the list of use cases, identified as MUST be addressed by XML 
>content in companion files:
>
>     * Users want to maintain standard WebCGM Application Structures (APS) 
> attributes in XML files
>     * Users identify WebCGM objects either by their unique APS 'id' or 
> non-unique 'name' attribute. Hence they want to associate XML data using 
> the 'id' or 'name', whatever is important to them.
>     * Current CGM users are limited by the fact that attributes and their 
> semantics are specified in the respective profiles.

So it is not presently written down in the 2.0 requirements.  Further 
consideration of this use case could be postponed till after 2.0.  But we 
need to be sure that this is not a critical requirement for one of our user 
("cascading") communities.

At least some TC members remember the final resolution, after some 
back-and-forth starting at Cologne, that *full* metadata enumeration is NOT 
a principal use case.

Regards,
-Lofton.