Re: comment on the syntax document: discrepancies between tests and the latest version...

Hi Ivan,

> I try to run my newest version of tests against crazy ivan, and I get
> four failed tests. As far as I can see, all four are based on some
> discrepancies between the syntax document and the test suite (or I
> misunderstood something).

Good tests...thanks.

> Here they are:
> Tests #34 (and #38 which is more or less the same)
> --------------------------------------------------
> In Test #34 we have
> <img    about=""
>                         rel="foaf:img"
>                         src=""
>                         alt="A photo depicting Michael" />
> and the requested triple is
> <> <>
> <>
> The current syntax document, however, uses @src as setting the subject,
> not the object. I know this was one of our evergreen discussion... what
> is it now?

Yep...@src sets the subject. So this example wouldn't actually create
any triples, but it would create a hanging @rel (although nothing
could make use of it).

We should probably tweak the test to this:

  <div about="" rel="foaf:img">
    <img src=""
      alt="A photo depicting Michael" />

or this:

      rev="foaf:img" resource=""
    alt="A photo depicting Michael" />

(Or perhaps both.)

Note that the big deal about using @src as a subject rather than an
object when @rel/@rev is present is first, that you can set the type:

      rev="foaf:img" resource=""
    alt="A photo depicting Michael" />

and second, you can make two statements about the image in one go:

  <div about="" rel="foaf:img">
    <img src=""
        rel="license" resource=""
      alt="A photo depicting Michael" />

There is no way to do this if @src is used as an object.

> Test #63
> --------
> What is the latest agreement on what the value of ":next" should be? It
> is not clear (to me) in the text. The test suggests that ":next" should
> be in the xhtml vocab space, but my fear is that this may be a previous
> version of our status (ie, the test should be changed)
> My understanding is that ":next" has an empty prefix, which should be
> the base URI...

Ok...we now have two-steps to the logic, and I apologise if the spec
is not clear. The first step is that @rel is defined as follows:

  linktype ::= 'next' | 'prev' | ...
  rel ::= (linktype | curie)*

The processing rules say that the link type maps directly to a other words, it has absolutely nothing to do with CURIEs.
This is 'logical', because that is exactly what we do with @about:

  about ::= URI | safe_curie

I.e., if @about holds a URI it is also nothing to do with CURIE processing.

So that takes care of @rel values that are valid link types. The
second step of our 'two-step logic' is that CURIEs with no prefix are
simply ignored. Note that the 'blank prefix' is a valid prefix, so
what this essentially means is that CURIEs with no colon in are

In other words, it's the second step that prevents values like
@rel="foo" from generating a triple.

In all of this, the default prefix is set to the XHTML vocabulary URI.
So @rel=":next" would be equivalent to @rel="next".

> Test #66:
> ---------
>   <head instanceof="foaf:Document">
>        <title>Test 0066</title>
>     </head>
>     <body>
>        <p>This is test #66.</p>
>     </body>
> The test suggests that
> <>
> <>
> <>
> ie, there is an implicit extra about on the <head>. I know we discussed
> this, we more or less agreed on that, but it is not on the syntax
> document as far as I can see...

I think you are apologies again.



  Mark Birbeck | +44 (0) 20 7689 9232 | Ltd. is registered in England and Wales, number 03730711
  The registered office is at:

    2nd Floor
    Titchfield House
    69-85 Tabernacle Street
    EC2A 4RR

Received on Thursday, 31 January 2008 11:21:56 UTC