Re: What is normative?

+1 Annette!

2015-05-18 18:07 GMT-03:00 Annette Greiner <amgreiner@lbl.gov>:

> Take a look at how WCAG does it [1]. They don’t use RFC2119 keywords.
> Instead, they add “(Level A)”, “(Level AA)”, etc.  Where terms like “must”
> and “should” arise in the text, they are treated as they are used in plain
> English. That obviates the awkwardness of trying to make keywords that were
> developed for specifying a technology work for best practices documents. In
> my view, using RFC2119 keywords makes our documents appear to be imposing
> actual requirements, which I think is potentially confusing for readers. It
> suggests that failure to follow a given BP will prevent users from being
> able to access the data. The nice thing about going without the keywords is
> that it means people can claim a lower level of conformance and still feel
> good, whereas people who meet the higher standard can claim that and feel
> even better. We wouldn't need to compromise on what we expect from people,
> and we could provide some stretch goals.
>
> One particular section of that RFC particularly bothers me in considering
> its use for best practices. It’s the following:
>
> 6. Guidance in the use of these Imperatives
>
>    Imperatives of the type defined in this memo must be used with care
>    and sparingly.  In particular, they MUST only be used where it is
>    actually required for interoperation or to limit behavior which has
>    potential for causing harm (e.g., limiting retransmisssions)  For
>    example, they must not be used to try to impose a particular method
>    on implementors where the method is not required for
>    interoperability. [2]
>
> [1] http://www.w3.org/TR/WCAG20/
> [2] https://www.ietf.org/rfc/rfc2119.txt
>
> --
> Annette Greiner
> NERSC Data and Analytics Services
> Lawrence Berkeley National Laboratory
> 510-495-2935
>
> On May 18, 2015, at 12:09 PM, Joao Paulo Almeida <jpalmeida@ieee.org>
> wrote:
>
> Hi Annette,
>
> That just changes the use of the normative statements a bit.
>
> I proposed to interpret the normative statements in the following way: if
> you claim conformance, you MUST, ...
>
> What you are proposing sounds like: if you claim conformance to level X,
> you MUST, ...
>
> regards,
> João Paulo
>
>
>
> On Mon, May 18, 2015 at 3:32 PM, Annette Greiner <amgreiner@lbl.gov>
> wrote:
>
>> We’ve had an idea at various times to assign a rating system, something
>> like the five stars but different enough to avoid confusion. I still think
>> that’s the best way to deal with this issue. It enables a publisher of data
>> to claim a concrete level of compliance, much like the WCAG.
>> -Annette
>> --
>> Annette Greiner
>> NERSC Data and Analytics Services
>> Lawrence Berkeley National Laboratory
>> 510-495-2935
>>
>> On May 18, 2015, at 8:17 AM, Phil Archer <phila@w3.org> wrote:
>>
>> > The issue is open in tracker so I'm taking it as open - but if we're
>> taking them out (and I think we are too) then some of the intro matter and
>> the template need updating.
>> >
>> > Phil
>> >
>> > On 18/05/2015 16:03, yaso@nic.br wrote:
>> >> I thought we had an agreement on this:
>> >>
>> >> "An alternative would be not to include any RFC2119 keywords at all"
>> >>
>> >> I ran trough the logs and couldn't find nothing against not using the
>> >> RFC2119 keywords at the document. Furthermore, we talked at the F2F
>> >> about the translation to Portuguese problem with the keywords. There
>> was
>> >> another decision on that?
>> >>
>> >>
>> >> yaso
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >> On 05/18/2015 11:53 AM, Phil Archer wrote:
>> >>> Dear all,
>> >>>
>> >>> The BP editors have been working hard and have made a number of what I
>> >>> think are big steps forward with the doc.
>> >>>
>> >>> But Issue-146 remains unresolved: what is normative in a BP?
>> >>>
>> >>> Take our old favourite first BP
>> >>> http://w3c.github.io/dwbp/bp.html#ProvideMetadata that says:
>> >>>
>> >>> Metadata MUST be provided for both human users and computer
>> applications
>> >>>
>> >>> I doubt anyone here will disagree with this statement, but is it right
>> >>> to make this the normative part of the BP? And, if so, are we right to
>> >>> use the RFC2119 MUST?
>> >>>
>> >>> Take a less clear cut example:
>> >>> http://w3c.github.io/dwbp/bp.html#MultipleFormats that says:
>> >>>
>> >>> Data SHOULD be available in multiple data formats.
>> >>>
>> >>> Really?
>> >>>
>> >>> SHOULD is "comply or explain" - i.e. you'd better have a very good
>> >>> reason not to provide data in multiple formats so I might argue one
>> day
>> >>> that this should be a MAY. What does MAY mean? From the infamous
>> RFC2119:
>> >>>
>> >>> "This word, or the adjective "OPTIONAL", mean that an item is
>> >>>    truly optional.  One vendor may choose to include the item because
>> a
>> >>>    particular marketplace requires it or because the vendor feels that
>> >>>    it enhances the product while another vendor may omit the same
>> item."
>> >>>
>> >>> (I've omitted the rest of the definition but this is the essence of
>> it).
>> >>>
>> >>> Suppose the WG agrees and this BP now becomes:
>> >>>
>> >>> "Data MAY be available in multiple data formats."
>> >>>
>> >>> Which doesn't really convey in a single sentence what we mean. We
>> might
>> >>> end up with
>> >>>
>> >>> "Publishers are encouraged to make data available in multiple formats
>> >>> (OPTIONAL)"
>> >>>
>> >>> i.e. re-word the normative line to fit in with the definition of the
>> >>> relevant RFC2119 keyword.
>> >>>
>> >>> An alternative would be not to include any RFC2119 keywords at all.
>> I'm
>> >>> easy either way - I can see arguments for and against including these
>> >>> keywords - but it remains an open issue that I think we owe it to the
>> >>> editors to decide what to do.
>> >>>
>> >>> Phil.
>> >>>
>> >>>
>> >>>
>> >>>
>> >>
>> >
>> > --
>> >
>> >
>> > Phil Archer
>> > W3C Data Activity Lead
>> > http://www.w3.org/2013/data/
>> >
>> > http://philarcher.org
>> > +44 (0)7887 767755
>> > @philarcher1
>> >
>>
>>
>>
>
>


-- 
Bernadette Farias Lóscio
Centro de Informática
Universidade Federal de Pernambuco - UFPE, Brazil
----------------------------------------------------------------------------