Re: editorial review of TriG

On Wed, Sep 18, 2013 at 6:58 AM, Sandro Hawke <sandro@w3.org> wrote:

>  Excellent.
>
> Two remaining typos and one new one, plus a couple in-line side comments
> (not critical for Last Call).
>
> 1.   "are considered to be part of the deafult graph"
>
> (spelling of default)
>

... ran spellcheck...


>
> 2.   Security Considerations
>
> "may also be used on Turtle documents"
> "retrieved from untrusted Turtle documents "
>
> Those should both be TriG, I believe.
>

Yes.


>
> And then some inline comments:
>
>
> On 09/16/2013 05:56 PM, Gavin Carothers wrote:
>
>
>
> On Sun, Sep 15, 2013 at 8:21 PM, Sandro Hawke <sandro@w3.org> wrote:
>
>>  Here are the editorial issues I found in the current TriG editor's draft
>> https://dvcs.w3.org/hg/rdf/raw-file/3fcaf6781dc8/trig/index.html
>>
>> None seems serious, and hopefully all can be fixed before publication.
>>
>> I have a few larger questions which I'll send separately, and I'm going
>> to try to do an implementation to check the grammar and test suite.
>>
>>     -- Sandro
>> Status of This Document We should include a statement here to the effect
>> that we might skip CR, and pointing folks at the test suite, implementation
>> report, etc.  For example see the red box in the Status section of
>> http://www.w3.org/TR/2012/WD-sparql11-query-20120724/
>>
>> We can do this by hand during publication if respec somehow makes it hard.
>> 2. An Introduction to TriG
>>
>> The IRI or blank node label of the graph statement may be used in another
>> graph statement which implies taking the union of the tripes generated by
>> each graph statement.
>>
>>
>> That's a pretty confusing sentence for me, and I know what it's trying to
>> say.   Plus, if we're being pedantic, the grammar also allows
>> blankNodePropertyLists and Collections to be used as graph names.    Maybe
>> go up a level and just say a trig file is the same as turtle, except there
>> are additional graph statements, each of which pairs a term (which is
>> evaluated to form a graph name) and a graph in { }.    Or something like
>> that.
>>
>
>  It does not allow for  blankNodePropertyLists and Collections to be
> graph labels.
>
>
>> 2.1 Graph Statements
>>
>>  The graph name of a graph statement may be omitted.
>>
>> That sounds confusingly like one could say "GRAPH { ... }".     Maybe we
>> can change things to not call default graph triples "graph statements"?
>>
>
>  All of the explanatory text was written before bare triples were legal.
> I need to revise most of the explanatory text clearly since it's no longer
> clear. Text updated.
>
>  One CAN say { ... }
>
>
>>  A RDF Dataset may contain only a single graph.
>>
>> Could be read as saying a dataset is only allowed to contain one graph.
>> Perhaps change to, "A simple RDF Dataset might contain only a single graph
>> statement".
>>
>
>  s/may/might
>
>  Done.
>
>
>>
>> 4.4 Escape Sequences In the table, the heading "Strings" should probably
>> be a link like the heading local_names
>>
>>
>>
>  Done.
>
>>  5. Parsing
>>
>> This section maps a string conforming to the grammar in section Not
>> found 'sec-grammar-grammar'<https://dvcs.w3.org/hg/rdf/raw-file/3fcaf6781dc8/trig/index.html#sec-grammar-grammar>to a set of triples by
>>
>> typo in section name, I assume
>>
>
>  Fixed.
>
>
>>
>> 5.1 Parser State I love the idea of linking to tests here, but alas,
>> none of those links are working.  I guess the links were copied from turtle
>> as relative links.  (Which is ironic, since the first tests are for @base.)
>>
>
>  This is why my action to replace the text at /TR/Turtle/tests/ has taken
> so long, I have no idea what to do with these to avoid 404s when the tests
> should not have been published here to start with.
>
>  Links now go to  https://dvcs.w3.org/hg/rdf/file/default/trig/tests/
>
>
> *shrug* I guess that's okay, although I'm a bit skeptical that link will
> still work in 5 years.
>
> We have to be very careful not to copy the tests over into /TR during
> publication.    (That is, I have to not copy it, if I do the copying, and
> you have to make sure the Webmaster doesn't, if you do the publication
> without me.)
>

Removed links to tests, they don't match anyway, and the fragments don't
work.


>
>
>
>> When undefines, triples are detined for the default graph.
>>
>>  typo in "undefines" and "detined"
>>
>
>  Fixed.
>
>
>>
>> I'm not seeing any text about what this is supposed to generate:
>>
>>    PREFIX : <http://a.example.org/> <http://a.example.org/>
>>    PREFIX : <http://b.example.org/> <http://b.example.org/>
>>    :c :c :c.
>>
>> Actually, looking in the Turtle spec, I don't see anything like that
>> there, either.
>>
>
>  Eh? One Triple:
>
>  <http://b.example.org/ <http://b.example.org/>c> <http://b.example.org/<http://b.example.org/>
> c> <http://b.example.org/ <http://b.example.org/>c> .
>
>
>    - Map[prefix <http://localhost/%7Egavin/rdf-wg/rdf-turtle/#prefix> ->
>    IRI] namespaces — The second and third rule arguments (PNAME_NS and
>    IRIREF) in the prefixID production<http://localhost/%7Egavin/rdf-wg/rdf-turtle/#grammar-production-prefixID> assign
>    a namespace name (IRIREF) for the prefix (PNAME_NS). Outside of a
>    prefixID production, any PNAME_NS is substituted with the namespace
>    (test: prefix1<http://localhost/%7Egavin/rdf-wg/rdf-turtle/tests/#prefix1>
>     escapedNamespace1<http://localhost/%7Egavin/rdf-wg/rdf-turtle/tests/#escapedNamespace1>
>    ). Note that the prefix may be an empty string, per the PNAME_NS,
>     production: (PN_PREFIX)? ":" (test: default1<http://localhost/%7Egavin/rdf-wg/rdf-turtle/tests/#default1>
>    ).
>
>
>
>
> I don't think that text makes is clear whether repeating the same prefix
> is allowed in Turtle/TriG, and if it's repeated if the namespace is allowed
> to be different.     I suppose this is a Turtle issue, not a TriG one.   Do
> you happen to know if we have test cases on this?
>

Yes.
http://www.w3.org/2013/TurtleTests/prefix_reassigned_and_used.ttl



>
>
>
>         -- Sandro
>
>
>
>> 5.3 RDF Triples Construction
>>
>> has some @@'s!
>>  5.3.1 Output Graph
>>
>>
>> I find the use of "unset" and "undefined" a little unsettling.   I think
>> it would be clearer to have a special value for the default graph.  But
>> whatever.
>>
>  There is a special value it's called null, nil, None ;)
>
>
>>   The rule labelOrSubject<https://dvcs.w3.org/hg/rdf/raw-file/3fcaf6781dc8/trig/index.html#grammar-production-labelOrSubject>sets both
>> curGraph. and curSubject
>>
>> extra period
>>
>
>   Fixed.
>
>>  5.4 Differences from Previous TriG
>>
>> This section describes the main differences between TriG, as defined in
>> this ddcument, and the
>>
>> Typo 'ddcument' and the sentence just ends in the middle.
>>
>
>  Ah yes, I wasn't sure what to call the old TriG, it's not a Team
> Submission, or a Member Submission...
>
>  "This section describes the main differences between TriG, as
> defined in this document, and earlier forms"
>
>>
>> 5.5 Internet Media Type, File Extension and Macintosh File Type
>>
>> I'd suggest just calling this Media Type Registration
>>
>
>  Done.
>
>
>>  This information that follows has been submitted to the IESG<http://lists.w3.org/Archives/Public/www-archive/2007Dec/0065.html>for review, approval, and registration with IANA.
>>
>> that's for Turtle.   Instead for TriG say it will be submitted.
>>
>> In Security considerations, the formatting is weird -- there should be
>> blank lines between paragraphs, I guess.  And there are still a few
>> references to turtle that should be trig.
>>
> They aren't paragraphs, they are dt dds, the styling and format comes from
> the default w3c stylesheet.
>
>
>>   In Base URI, I'd say "The TriG base directive can change...." so that
>> it glosses over BASE vs @base.
>>
> Done
>
>>   That's it!
>>
>
>
>  Done!
>
>
>