Re: LC on Test and Primer - timescale issues

DanC:
> I lean toward going to PR by WWW2007 with whatever
> we have in hand. For details, again, see
>   http://www.w3.org/2001/sw/grddl-wg/sched7
>   <- http://www.w3.org/2001/sw/grddl-wg/#sched
> 
>>  Clearly, we will have a better set of 
>> deliverables if we do, and as long as their is sufficient commitment 
>> from the editors to do the necessary work, then I am supportive.
>>

Summary:
- advocacy of WG Note status for Primer and Use Cases
- advocacy for getting on with the test work


In light of DanC comments, and in light of HP's hope that he will 
succeed in the new HTML role, I remain concerned that the publication 
ambitions of the group will cause unnecessary delay.

Background
==========

My understanding is that the primary goal of this group is to advance 
the GRDDL Spec to PR (without any issues that may prevent it from then 
being advanced by the director to Rec).

For this we need:

A) a high enough quality LC version of the GRDDL Spec.
B) sufficient evidence that it can be implemented.
C) to complete the last call process on the GRDDL Spec

We have (A); we don't have (B) - although we know that it can be 
implemented, we can't adequately document this fact.

Thus work on the tests and the test document is on the critical path.

Primer and Use Cases
====================

Work on the primer and the use cases document is not on that critical 
path. This group could succeed with no further work, at all, on these 
documents. Obviously we prefer the success to be better than that, and 
hence advancing the use cases and the primer to some satisfactory 
end-state is desirable.

In the WG Note versus Rec decision for both documents (Use Cases and 
Primer).

+ A Rec is likely to be a better document: it will generally have had 
wider review.
+ A Rec is a better entry on the editors' CV
+ A Rec is a better advertisement for the member who is paying for the 
editors' time.
+ A Rec is significantly more work for the editor.
+ A Rec takes longer.
+ A Rec is more work for the working group
+ A Rec is more work for the team contact

I generally feel that when the editors are willing to do the additional 
work, in a timely way, then the improved quality of the documents is 
worth the extra work for the working group.

In this case, we seem to be under significant time pressure, partly due 
to DanC's commitments, and to the justified expectation of W3C that we 
will finish in a timely way, given that this WG has not had any really 
difficult issues of the sort that derail a group's timeline.

Without an adequate timeline, with commitment from the editors to that 
timeline, then I oppose rec-tracking either of these documents.

With the use cases in particular, this document is most useful, during 
the work of the working group: the benefit of quality improvements 
during a last call process is likely to be minimal. (But also the cost 
of rec tracking use cases is not high)

With the primer, I am concerned that there is a backlog of comments 
needing to be added. This demonstrates a lack of editorial commitment to 
the group's timeline. Hence I prefer to take the lower risk, lower cost, 
and (slightly) lower benefit option of taking that along the WG Note route.

Test
====

For advancing to PR we need to be able to document that the spec is 
implementable.
A good way to do this is to receive implementation reports from multiple 
implementations, that indicate sufficient passes of sufficient tests of 
the spec.

This requires a test suite that is sufficiently clear, correct and easy 
to use, and sufficiently conformant with the spec.
1) A good way to achieve this is a rec track document.
2) Another way to achieve this a WG Note document.
3) A third way to achieve this is simply to put enough effort into the 
test suite.

I understand that we have already decided for (1) or (2), and against (3).

Option (1) is harder than option (2), which is harder than option (3).

The quality of results from option (1) is likely to be better than with 
option (2), is likely to be better than with option (3).

As I pointed out earlier, option (1) requires a first Test WD very very 
soon; or a change in our timeline.
Option (2) can be achieved with a first WD in say about a month, and 
then the final version being a WG Note, when the spec moves to PR.

I don't believe option (3) offers enough incentive for anyone to do the 
necessary work.

Jeremy

Received on Tuesday, 13 March 2007 11:27:35 UTC