> From: Andy Powell [mailto:[log in to unmask]]
> Sent: Wednesday, July 23, 2003 10:56 PM
> To: Julian Reschke
> Cc: [log in to unmask]
> Subject: Re: Implementing according to RFC2731
> On Wed, 23 Jul 2003, Julian Reschke wrote:
> > we're currently working on extracing data from HTML based on RFC2731.
> > main issues we encountered so far are:
> I'm not sure that you'll get any firm answers to these questions... but
> I'll give you my personal thoughts.
OK, thanks for the initial feedback...
> > - when processing the HEAD part, is ordering of LINK and META tags
> > Is it ok if the link element mapping the schema prefix to a URI occurs
> > *after* the META tag using it? What is the scope? Is it allowed to have
> > multiple link elements that map the same prefix to different URIs?
> I would think that the ordering in not significant - and that therefore
> you should not have multiple link elements that map the same prefix to
> different URIs.
However that makes stream processing of the HEAD content a bit harder
(because you need to delay processing of the META tags until all LINK tags
have been seen). For <http://www.ukoln.ac.uk/metadata/dcmi/dcq-html/>,
wouldn't it be much simpler to define the scope of the LINK schema
declarations as "all following content", such that things like
<meta name="dc:date" scheme="dcterms:W3CDTF" content="2001-07-18" />
<link rel="schema.dc" href="http://purl.org/dc/elements/1.1/" />
become illegal? Or are there really concerns about backwards compatibility
with existing documents and specs?
> > - uppercase/lowercase: similar to WebDAV (RFC2518), our system
> > properties based on a namespace name (URI reference) and a
> local name (XML
> > element name). In WebDAV, case *is* relevant. What's the
> convention for DCMI
> > properties?
> The convention in HTML has tended to be for element names to start with an
> uppercase letter (DC.Title). In other syntaxes (i.e. XML and RDF/XML)
> that make proper use of XML namespaces, all element names start with a
> lowercase letter (dc:title). That's why in the draft DC/HTML
> recommendation at
> I recommend that applications should always generate lowercase first
> letters, but accept letters in any (mixed) case.
I think this is not good enough. I may want to include non-NC properties
into an (X)HTML document where the property names *are* case-sensitive.
Would it possibly make sense to define that
- if the schema prefix separator is a dot (RFC2731), the propertyname should
be lower-cased, and
- otherwise (":"), the property name *is* case-sensitive?
Related: is the case of the schema prefix significant? I think when using
the ":" notation it SHOULD be for the sake of consistency with similar usage
in W3C specs.
> > - assuming that I'd want to use RFC2731-style encoding to map
> > properties into my own (case-sensitive) property schema -- is there any
> > reliable way to find out whether for a given scheme (such as
> > "http://purl.org/DC/elements/1.0/") case is relevant or not?
> > - versioning: do "http://purl.org/DC/elements/1.0/" (RFC2731),
> > "http://purl.org/dc/elements/1.1/"
> > (<http://www.ukoln.ac.uk/metadata/dcmi/dcq-html/>) and
> >(example usage in the W3C home page) identify the same property schemas?
> is the namespace URI for version 1.1 of the DC elements.
OK, I found <http://dublincore.org/documents/dcmi-namespace/> which
clarifies a lot. I guess I'll ask the W3C to fix their usage of DC.*
elements in their META tags then :-)
> I'm not sure what the other two URIs you give actually define. However,
> my feeling is that the semantics of the 15 DC elements changed
> sufficiently little between version 1.0 and version 1.1 that you can
> reasonably treat the 3 URIs you cite above as being equivalent in the
> context of HTML link tags.
<green/>bytes GmbH -- http://www.greenbytes.de -- tel:+492512807760