HTTP RDF Update review

This is my review of http://www.w3.org/2009/sparql/docs/http-rdf-update/ 
in discharge of action-167.

Overall, I support publication of th document as-is as a next Working Draft.

~~~ general comments ~~~

* The introduction needs to explain the relationship (or lack thereof) 
between this and the SPARQL Protocol. This will also be helped if we 
assemble an overall SPARQL guide doc, but we don't have that yet.

* Terminology. I think this section is well thought-out in terms of its 
precision, but nevertheless is difficult to take in as a reader of the 
specification. I wonder if we could move this to later in the document 
and hyperlink the terms when they occur in the context of the specification?

* I appreciate the lengths the document goes to to correctly interpret 
both the semantics of HTTP and of RDF. I think, though, that the 
document would benefit from presenting the most straightforward cases in 
a straightforward manner. Perhaps a section with examples that consist 
of setup (e.g. "a graph store stores graphs g1 and g2 and implements 
this protocol at URI u1"), request, response, and effect would be 
useful. These examples could also help the reader with the terminology, 
by making clear, e.g., what exactly is the networked-RDF knowledge in a 
particular example.


~~~ minor comments ~~~

* "This protocol specifies HTTP operations for managing 
network-manipulable RDF datasets as well as their semantics". Unclear 
what "their semantics" refers to. Suggest "This protocol specifies the 
semantics of HTTP operations for managing network-manipulable RDF datasets"

* In section 3, shouldn't the example request only have the URI path in 
the GET line?

    GET /rdf-graphs/employees HTTP/1.1
    Host: example.com
    Accept: application/rdf+xml

If so, the text following the request needs to also be changed to 
reflect this.

* Section 3 - in FF3.6 on Windows, the image appears inline rather than 
below the text that references it.

* 4.1  "SHOULD can be used" -> "SHOULD be used"

* 4.1 I don't understand what "identified facts" refers to. In any case, 
I think it's not necessary to show shorthand SPARQL Update statements. 
It's most straightforward to give one SPARQL Update translation of each 
HTTP operation.

* 4.3 Is "subordinate" a standard term? If so, can it be referenced? If 
not, it's meaning in this context is not clear to me.

* 4.3 "usecase" -> "use case" ... but I don't think we need to explain 
the use case for POST, just what it is.

* "4.3 HTTP GET" -> "4.4 HTTP GET"

* "/o" -> "?o"


Lee

Received on Wednesday, 6 January 2010 08:15:50 UTC