RE: w3process-ISSUE-38 (documenting implementation): Add a should requirement to document known implementation [Document life cycle (ch 7)]

> Despite your comments, the intent of the requirement was simply
> to list known implementations.

I think I understand the current requirement and Charles' suggestion
that the current requirement be extended to all phases of development
and not just LCCR/Rec.

>  Various WGs have also chosen to show the how known implementations 
> fare with the current test available for a given application, but that is not a 
> requirement. 

It isn't currently a requirement in the current process.

But this is the "W3c-Process"  and I am suggesting a change to the
requirements which, I believe, will be less work, less controversial,
and more useful.  

Knowing whether to count an implementation is hard. Something
that implements 10% of the features, something that implements
90% of the features, which should count?

Making the working group evaluate implementation reports
requires quite a bit of caution and review, especially if the granularity
is just "yes, this is an implementation" vs. "no, it is not".

I want to short-circuit independent evaluation of implementations
by letting implementations self-report, rather than requiring the
working group to come to consensus on the list of known implementations.


Requiring that the list be maintained at all stages of development,
not just at LCCR/Rec, seems to me to add significant work
without corresponding benefit, because of the difficulty
of deciding when to count an implementation, especially
as the specification is undergoing rapid changes.

 But independent implementation reports can be gathered
asynchronously and maintained independently without the
working group having to manage it.



> > http://www.w3.org/community/w3process/track/issues/38

> >
> > Raised by: Charles McCathie Nevile
> > On product: Document life cycle (ch 7)
> >
> > There is currently a should requirement to document known
> > implementation at LCCR / Rec. It actually seems like it should be a general
> requirement.
> >
> >

Received on Monday, 12 August 2013 10:20:39 UTC