Re: Review of "SPARQL 1.1 RDF Dataset HTTP Protocol"

Minor points - nothing for LC:


>> [*] "*MUST* ....and the words appear as emphasized text, "
>>
>> The document does not use them in this way. They are not in bold later, and
>> bold is reserved for HTTP Verbs.
>> SmallCaps might be useful to emphasize.
>
> I switched to italics rather than bold.

What's the intention for the rest of the document?  These words are not 
used italized.  I was expecting the RFC 2119 text to be in the same 
style as used in the document as a whole: capitals would be the easiest way.

e.g.
"""
A compliant implementation of this specification SHOULD accept HTTP 
requests directed at its Graph Store
"""
has SHOULD in plain upper case.

(not necessary for LC)

>> [] Serialize (verb)
>> Does not work for me.
>>
>> A document is a serialization of a graph (noun)
>>
>> An alternative use of the verb Serialize is the action of doing it, and
>> that's done by the application code/CPU.
>>
>> "A document serializes graph." does not seem right to me if read as the
>> document is carrying out the action.  In fact, the document avoids such a
>> phrase and says:
>>
>> "document ... which serializes ... graph."
>> "document ... that serializes ... graph."
>> which isn't a simple subject-verb-object sentence.
>> and "graph serialized by document" occurs as well.
>
> Ok.  I added this term in response to Ian's comment about being
> unfamiliar with using serialization and representation
> interchangeably.  So, I was trying to spell out the use of the word
> serialization in relation to a graph and an RDF document.  In my
> opinion, there is enough precedent in using this term in the sense
> that I meant both from the RDF/XML specification ("An RDF Document is
> a serialization of an RDF Graph into a concrete syntax.") and the
> original SPARQL specification ("the resource is represented by a [..]
> a document that serializes a graph").  However, there seems to be a
> significant amount of confusion regarding the idea that a graph URI
> identifies a resource that is represented by a document that
> serializes the corresponding graph (despite the fact that this idea is
> part of the SPARQL 1.0 specification).
>
> So, I will remove this term definition, however, I wouldn't be
> surprised if this continues to remain a point of confusion.

Fine - if there is to be a defintion, make like your example, which is 
not a verb.

"An RDF Document is
a serialization of an RDF Graph into a concrete syntax."

>> [**] "compliant"
>> There is no compliance section in the document.
>
> See my question above about whether compliance sections are required
> for all W3C REC track documents.

Yes - it's a WG-wide matter.

>> [*] Suggestion: Add example the HTTP operation before the SPARQL Update
>> equivalent.  Speaks to KK comment about implementers.
>
> In every case where there is a SPARQL Update equivalent, I have added
> an HTTP example before hand.

Excellent - I think it works well this way.

	Andy

Received on Sunday, 6 February 2011 15:19:16 UTC