Re: N-Triples - Please encourage ntriples tools to emit *canonical* N-Triples

On 12/06/2013 05:19 AM, Andy Seaborne wrote:
> (not an official response)
>
> David,
>
> It is too late in the spec cycle to introduce SHOULD/MUST language -- it
> is asking implementers to read the spec differently to the LC, CR versions.
>
> What I will suggest to the WG is making the description of canonical
> N-Triples a separate section, not just a subpart of the conformance
> section.

I think that's an excellent idea.

David

>
> While this is "editorial" - it's the same text, just moved and reworded
> if it didn't read correctly any more - we are very close to the
> transition to PR so I feel it needs to be discussed by the WG.
>
> Draft in editors working draft.
>
>      Andy
>
> On 03/12/13 22:41, David Booth wrote:
>> Regarding section 4, Conformance:
>> https://dvcs.w3.org/hg/rdf/raw-file/default/rdf-turtle/n-triples.html
>>
>> It would be helpful if tools that generate N-Triples were strongly
>> encouraged to generate it in canonical form unless there would be a
>> significant loss to the user in doing so.  For example, if a tool
>> normally generates useful information in comments, it may be better to
>> generate non-canonical N-Triples, since canonical N-Triples would
>> require the comments to be removed.
>>
>> I suggest adding something like the following to section 4:
>>
>>    "Tools that generate N-Triples documents SHOULD generate
>>    **canonical N-Triples documents** unless doing so would
>>    result in a significant loss in functionality or performance.
>>    For example, if a tool normally generates useful information
>>    in comments, it may be preferable to generate non-canonical
>>    N-Triples, since canonical N-Triples requires comments to be
>>    removed."
>>
>> Please note that the RDF 2119 definition of SHOULD is:
>> http://www.ietf.org/rfc/rfc2119.txt
>> [[
>> 3. SHOULD   This word, or the adjective "RECOMMENDED", mean that there
>>     may exist valid reasons in particular circumstances to ignore a
>>     particular item, but the full implications must be understood and
>>     carefully weighed before choosing a different course.
>> ]]
>>
>> Because SHOULD allows this judgement-call-based wiggle room, this change
>> would not have to affect any conformance tests.
>>
>> If the working group decides that a "SHOULD" would be too strong, please
>> instead add an editorial comment to the above effect instead, such as by
>> saying "are strongly encouraged to" instead of "SHOULD".
>>
>> Thanks,
>> David
>>
>
>
>
>
>

Received on Friday, 6 December 2013 21:34:03 UTC