Re: [SpecGL-impl] xml:id Last Call

At 03:09 PM 1/3/2005 -0500, Karl Dubost wrote:

>Le 02 janv. 2005, à 20:53, Lofton Henderson a écrit :
>>This Requirement derived from the earlier Checkpoint about "identify 
>>Classes of Product".  CoP were items for which the specification defines 
>>conformance requirements.
>>
>>I can't see how the referenced text in xml-id satisfies a requirement to 
>>identify classes of product.  This Introduction text only contains some 
>>vague  generalizations about what xml:id is for.  If it was our intent, 
>>in our attempt to avoid scary technical jargon, that xml:id passes the 
>>successor Requirement (2.2.A) to the original CoP requirement, then I 
>>think we have gone too far.
>
>My short answer: The best is the enemy of good.

My point:  I don't think this is good.  (this = the xml:id Introduction, as 
a satisfaction of the CoP requirement.)
http://www.w3.org/TR/2004/WD-xml-id-20041109/#intro


>My long answer: Editors and people in WG are humans not bots. I agree to 
>achieve the maximum of quality. And for myself I will always try to push 
>forward each bit of quality. To have worked in many environments where I 
>had to encourage people to create things, stuffs, etc. in the right way 
>without being constrained by let say a pay check, I can tell that 
>gratification is always better than the stick. The stick never works with 
>regards to that.

Let me be clearer.  I do not believe that xml:id satisfies our intention of 
the CoP requirement.  The QAWG needs to decide that.   Two possibilities,

1.) if QAWG thinks it satisfies the CoP requirement, then I think that 
requirement is far too weak and needs to be strengthened, more like it was 
in CR SpecGL.  Compare the xml:id Intro, for example, to SVG11 Confo clause:

http://www.w3.org/TR/SVG11/conform.html

2.) if QAWG thinks it does not satisfy the CoP requirement, then ...

Lecturing about humans and bots aside, I believe it is a disservice to 
xml:id (and everyone) to say "fine", when it is not fine.  How do humans 
ever improve, without constructive and helpfully critical feedback?  I 
would NOT like to hold up the xml:id Intro as an example of an unqualified 
YES to the CoP requirement,



>>IMO, what the referenced #intro text in xml:id says about conformance and 
>>Classes of Product is relatively worthless, and fairly obscure as 
>>well.  If it is our intent that xml:id passes that SpecGL requirement, 
>>then I think we have made the requirement much too wishy-washy (in other 
>>words, it is relatively worthless, IMO).
>
>Therefore I can understand you are not satisfied with my review and ask 
>for modification, that's perfectly fine and normal.  But more than 
>trashing the whole requirement, please improve it.

First things first, Karl.  I am not "trashing the whole requirement".  In 
fact, first I'm asking whether other QAWG think xml:id satisfies the 
requirement.  (I think the answer should be "no".)

>         Give the wording you think that will be better.

I actually think that the "what means", "why care", and "techniques" gives 
a fairly good indication of what we expect:
http://www.w3.org/TR/2004/WD-qaframe-spec-20041122/#implement-principle

If anything, the statement of the requirement is what (IMO) is too weak, 
and is somewhat misleading.  One might argue that xml:id satisfies the 
short statement of the requirement.  But I wouldn't say it satisfies the 
explained meaning, intent and techniques.  Isn't that what we intend to be 
satisfied?

-Lofton.

>         And even BETTER give the techniques and the template that will 
> help to achieve it. If you are not satisfied it's always better to come 
> with something that will illustrate what you are saying.
>         Another thing you could do is rewrite the Xinclude paragraph to 
> show what would have been the prose that will fulfill the requirement.

Received on Tuesday, 4 January 2005 16:14:23 UTC