- From: Joseph Reagle <reagle@w3.org>
- Date: Wed, 4 Sep 2002 12:54:21 -0400
- To: Ian Jacobs <ij@w3.org>
- Cc: www-tag@w3.org
http://www.w3.org/TR/2002/WD-webarch-20020830/ 2. Formats. A nonexclusive set of data format specifications designed for interchange between agents in the system. This includes several data formats used in isolation or in combination (e.g., XHTML, CSS, PNG, XLink, RDF, SMIL animation), as well as technologies for designing new data formats (XML, XML Namespaces). Is RDF rightfully considered a format? I think of it as a data-model that can be expressed in the XML format. (The use of "data format" to describe an application probably is correct, but counter to my intuition.) Editor's note: While people agree that URIs identify resources (per [RFC2396]), there is not yet consensus that absolute URI references with fragment identifies may be used to identify resources. Some people contend that an absolute URI reference with a fragment identifier identifies a portion of a representation. I don't know what the exact answer to the latter part of the paragraph is, but xmldsig/xenc certainly uses frag-id's as part of the identifier... 2.1. Resources, URIs, and the shared information space Use absolute URI references: All important resources SHOULD be identified by an absolute URI reference.1 What exceptions are permitted? Is this to accommodate the use of QNames as identifiers? (Can one have an identifier without a "resource"?) 2.2.1. Comparison of identifiers Issue: URIEquivalence-15: When are two URI variants considered equivalent? I raised this issue with the TAG a while back based on a question from Oasis folks. FYI: I thought SAML did a very specific character for character equivalence (to match between an accessed resource and the security policy corresponding to it), but I no longer can find appropriate text and the XACML work appears to permit a rather open interpretation. (However, I had a difficult time finding and reading though Oasis documents as they're in PDF, Word, and zip...) [[ http://www.oasis-open.org/committees/rights/#documents 9.11.Resource Matching A common example of this is a Web Server. Commercial http responders permit a variety of syntaxes to be treated equivalently. The “%” can be used to represent characters by hex value. In the URL path “/../” provides multiple ways of specifying the same value. Multiple character sets may be permitted and in some cases, the same printed character can be represented by different binary values. If the policy target matching algorithm considers two resource strings to be different, and the underlying Web server considers them to be the same, this may allow unintended access. The usual solution to this problem is to put the request in a canonical form before matching. There may be practical difficulties with this strategy if the transformations are not completely documented or subject to change without notice from one version to the next. It is important to be aware of this issue and perform careful checking of marginal cases ]] 2.2.4. Absolute URI references and context-sensitivity Each absolute URI reference unambiguously identifies one resource, but the resource itself may be defined in a context-sensitive manner. Is the actual language (e.g., content negotiation) considered a different representation of the same resource? 2.4. URI Schemes Do not use unregistered URI schemes: Unregistered URI schemes MUST NOT be used on the public Internet. I'd move this box below the following paragraph, "The IANA" because what it means to be a registered or unregegistered media type is not yet defined.
Received on Wednesday, 4 September 2002 12:54:28 UTC