Re: [CSS21] response to issue 15b

On 2/10/04 11:07 AM, "Bjoern Hoehrmann" <derhoermi@gmx.net> wrote:

> 
> * Ian Hickson wrote:
>> The other part states that features that have not been adequately tested
>> will be dropped. Are you seriously suggesting that you would want features
>> that have not received adequate interoperability testing to remain in the
>> specification? Isn't that violating the spirit of the process document
>> (and the whole _point_ of a CR stage) much more than the given criteria?
> 
> The Process document requires to precisely identify features considered
> at risk to ensure that the Proposed Recommendation does not invalidate
> an individual's review or implementation experience of the Candidate
> Recommendation.

That requirement has been satisfied.  The expression "any features not in
CSS1" is a precise set.

> A technical report should not advance within the
> Recommendation track without further work if the Working Group is not
> able to demonstrate that each feature of the technical report has been
> implemented.

Agreed.  Thus such features may also be dropped.

> The proposed exit criteria allow dropping features just for the sake of
> advancement. I do not want to say that this is the intent of the
> proposed exit criteria, my point is just that a Working Group should not
> have a blank check to do so.

Not a "blank check".  This only applies to features that were not in CSS1.

> If implementors have simply not gotten around to implement a perfectly
> fine section of the specification, the perfectly fine section should not
> get dropped, 

Such "perfectly fine sections" will be dropped from CSS2.1 and moved to the
appropriate CSS3 module(s).

> the Working Group should rather wait with its request for
> advancement;

Which is what the working group will do with the CSS3 modules.

> it may also request advancement without demonstrating
> implementation experience.

This is a very bad idea, and should be avoid by all working groups.

Specifications without demonstrated implementation experience have proven to
be very problematic, e.g. CSS2.

> The Working Group should encourage complete implementations of the
> technical report,

The CR draft will do so.


> the proposed exit criteria however do not do so;

That is the wrong place to state encouragement.


> it
> does not seem unreasonable for implementors to expect that the Working
> Group drops the features they do not implement and they would still
> conform to the specification.

No that is unreasonable, as there are several implementors implementing.
Only two implementors need to produce interoperable implementations in order
for the feature to remain.  And which two implementation those are can be
different from feature to feature.


> If it is forseeable that the exit criteria will not be met due to a
> specific feature, it would also be reasonable to go back to Last Call
> with the feature removed, demonstrate implementations and request
> advancement of the technical report, without losing time or considerable
> additional work.

The working group prefers to complete in CSS2.1 in a timely manner, and move
such features to CSS3.


> The difference is that reviewers would have a chance to
> object to the removal of the feature, as they would have if the exit
> criteria precisely identify features considered at risk, which is all I
> have asked for.

The list was precisely identified, "features that are not in CSS1".
Reviewers have had a chance to object.

At this point the working group considers this issue closed.

Thanks for your feedback.

Tantek Çelik
for the CSS WG

Received on Monday, 16 February 2004 13:04:57 UTC