Re: News Metadata Framework Requirements draft published

Hi Misha,

The document doesn't say where to send comments, feel free to forward  
to the appropriate mailing list.

Le 05-06-22 à 07:24, Misha Wolf a écrit :
> A second public draft of the News Metadata Framework Requirements
> specification has been issued today by the International Press
> Telecommunications Council (IPTC) for public comment and for
> discussion with other Standards bodies.
>
> The Requirements specification is available at:
>     http://www.iptc.org/dev/
>
> Comments on the Requirements are welcome.


This is a review and comment of
http://iptc.org/pdl.php?fn=DRAFT-NAR_1.0-spec-NMDF-BusReq_34.pdf

1. General comments:
     Often in your document, you are using RFC 2119 keywords to set  
the mandatory nature of the requirement, BUT at the same time the  
sentence on which the MUST, SHOULD applies is vague, or have  
something which is not testable.  Be sure to fix the language so  
there might be no ambiguities.
      That would be easier to have an HTML version with anchors to be  
able to precisely give references to part of the content.
     When you will develop the framework, I encourage you to use the  
QA Specification Guidelines. It will help you to organize the work  
and make your specification more usable.
     http://www.w3.org/TR/qaframe-spec/
     A checklist is available to help you to check if the technology  
respects simple criteria of quality.
     http://www.w3.org/TR/qaframe-spec/specgl-ics
     If you need help to understand them, feel free to contact me.

2. Simplicity
     I deeply agree with this requirement, but I'm not sure how you  
define what is simple and what is complex. This requirement is not  
testable by itself. Maybe better would be an implementability  
requirement. Given a feature, how many people have been able to  
implement the feature and/or to use in a interoperable way. If not,  
it means there's a problem.

3. Define Users
     [[[
     2.2 Ease of Use
     The solution SHOULD minimize technical challenges for users,  
taking advantage, where possible,  of technologies that are widely  
adopted.
     ]]]
What is a user? someone using a software implementing the framework?  
or a user of the specification (developer) implementing the  
specification. The first one is using a UI and then the ease of use  
is more a UI problem. The second one has to develop the program.

4. Examples
     [[[
     2.4  Examples (Agreed)
     The solution MUST be accompanied by a broad set of examples.
     ]]]
     Not testable :) because "broad" implies something which is  
vague. You could say, each feature of the language MUST include at  
least one example.

5. in 3.1 Metadata Classes
     "Maximum choices" + MUST: not testable

6. In 3.10.3.2  Relevance (Agreed)
     "1 to 100" is maybe a too strong requirement.
        What about people who wants a choice of keyword, a visual cue  
to set the relevance, etc.

7. in 3.10.3.4  Description (Agreed)
     [[[The solution MUST allow the provider to specify, for each  
metadata container, a free-text  description of the container’s value.
     ]]]
     What about a possibility of multi-lingual description?

8. In 4.6  Management, evolution and extensibility
     I would encourage as defined in SpecGL, the specification MUST  
address extensibility and defines the mechanisms for it.


-- 
Karl Dubost - http://www.w3.org/People/karl/
W3C Conformance Manager
*** Be Strict To Be Cool ***

Received on Wednesday, 22 June 2005 19:52:14 UTC