Re: syntax review (points for Dave, Eric, Brian)

On Mon, 7 Jul 2003 09:54:13 +0300
Jeremy Carroll <jjc@hpl.hp.com> wrote:

> 
> 
> This is a review of the draft syntax draft number 1.502
> 
> I will start with the points for Eric and Brian 
> 
> Eric:
> 
> In the SOTD it describes the changes from M&S.
> I draw this to your attention since SOTD belongs to you.
> A possible change would be to delete the para
> "This documents revises ..."
> and to move the following para into the introduction

I've written all the SOTD so far, for syntax.  It
is the last thing to write, I feel.

> Brian:
> 
> There were a few places where I did not like the wording.
> Unfortunately I haven't time to suggest better wording, so I wondered if you 
> would double check, and then if you agree with me that the current wording 
> should be improved please suggest to Dave.
> 
> These points were:
> 
> the last para of 6.1.4

Yes, that does read a bit odd.  Suggestion

FROM:
    The construction of RDF URI references from XML attributes can
    generate the same RDF URI references from different XML
    attributes.  This can cause ambiguity in the grammar when
    matching attribute events (such as when rdf:about and about XML
    attributes are both present).  Documents that have this are illegal.

TO:
    Different XML attribute events can generate the same RDF URI
    reference.  This can cause ambiguity in the grammar when matching
    attribute events such as both when rdf:about and about XML
    attributes are present.  Documents that have both the namespaced
    and non-namespaced attribute on the same element are illegal.

(I'm not making this change yet, just offering it)


> The word "must" in the last para of the dfefn of striong value in 6.1.8 and 
> 6.1.9

There are several uses of must in these paragraphs.  Are you
suggesting they should be MUST?

> Also the whiteSpace facet note at end of 6.1.9

I'm not the W3C XML Schema expert - I added this after a WG decision,
on http://www.w3.org/2001/sw/RDFCore/20030123-issues/#xmlsch-02
from your (Jeremy) proposal in:
  http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2003May/0097.html


> Dave
> ====
> 
> Here starts the review
> 
> 
> Suggest delete
> section 1 para 2 historical interest only, minimally change tense

True.  Done.

> 
> throughout check every occurerence of "Working Drfat" and update to
> "Proposed Recommendation" appropriately

For this document or refering to others?  I think this needs a WG
decision to publish things before I can do that.


> 2.1 para 3
> ... appending ... after ...
> suggest s/after/to/

I prefer after; it more clearly tells you which was first.

> 2.1 para 3 sentence 3
> INCORRECT?

This is:
  "RDF URI references identifying subject and object nodes can also
  be stored as XML attribute value or XML element names using QNames"

> Suggest
> RDF URI references identifying subject and object nodes can also be stored as 
> XML attribute values.
> (delete last phrase)

Yes, you are correct.  Done.

> 2.1 para 3. suggest s/value/values/ at very end of para

Yes. Done.

> 2.2 immediatly after figure
> 
> suggest
> s/The Figure 2/The left hand side of Figure 2/

Yes, that is better. Done.

> 
> Para before example 1
> suggest
> ... the sequence of three nodes and two predicate arcs on the left hand side 
> of Figure 2 ...

Done.

> 
> 2.8 suggest saying explicitly that xml:lang has no effect

Is it not that no external in-scope xml:lang (or xml:base etc)
has an effect rather than xml:lang has no effect in the XML literal
content?

No changes made.

(Incidently I fixed a problem with the N-Triples output for
example09.rdf - it now is the correct canonicalization, I checked
with ARP2).

> 2.9 example 10
> weakly suggest
> inserting a newline between int">123
> i.e.
> s/int">123/int"\n   >123/
> 
> so that the page is wide enough

I don't like that style personally, and it doesn't match the other
examples, so I'm not going to change this.


> 2.14
> I copied you on the spurious defect report on ARP concerning what to do when 
> no base URI for the document is known and no xml:base is given.
> It might be worth adding to this section that
> "Both forms require a base URI to be known, either from xml:base or for the 
> RDF/XML document." (probably could do with wordsmithing)

OK.  I've added:

  Both forms require a base URI to be known, either from an in-scope
  xml:base or from the URI of the RDF/XML document.

> 
> 2.16 para 1 
> suggest delete last sentence

The order of collections has come up several times and the
XML-derived syntax order has been a problem in the past (rdf:li).  I
felt it was better to flag this earlier.


> 4 Registration note still has a FIXME in it - but I don't understand this part
It isn't clear what RDF Core is going to do with the mime type
document at this time.  It is possible we might embed it in this
section (after TAG finding) or write a different commitment about the
RFC.

> 5.2 in contrast with 2.14 this is clear enough on base URIs
> 
> 6 first para - very good
> 
> 7.2.19
> suggest 
> s/is given, the above/is given, either of the above/

Done

> relaxng grammar not reviewed
> 
> in change list suggest spelling Duerst with a u-umlaut

Yes, I missed that (correct in refs).  Done.

> I support moving this document to PR

Thanks.

These changes are made in the ILRT CVS 1.503 at
  http://ilrt.org/discovery/2001/07/rdf-syntax-grammar/

Dave

Received on Monday, 14 July 2003 07:57:22 UTC