- From: Julian Reschke <julian.reschke@gmx.de>
- Date: Fri, 14 Mar 2008 14:04:55 +0100
- To: Mark Nottingham <mnot@mnot.net>
- CC: Harry Halpin <hhalpin@ibiblio.org>, ietf-http-wg@w3.org
Mark Nottingham wrote: > > I've submitted a draft-01 (perhaps against my best judgement; this > discussion is rapidly consuming the mailing list...), which should > appear shortly. In the meantime, it's available at > <http://www.mnot.net/drafts/draft-nottingham-http-link-header-01.txt>. > > It is in no way a finished product; it's just a straw-man that tries to > cover the issues, so it's easier to appreciate what they are (even if we > don't go in this direction). > ... Mark, I think this absolutely heads into the right direction. A few comments...: 1. Introduction A means of indicating the relationships between documents on the Web has been available for some time in HTML, and was considered as a HTTP header in [RFC2068], but removed from [RFC2616], due to a lack of implementation experience. JRE: include reference for HTML relationship = URI-Reference | <"> URI-Reference *( SP URI-Reference) <"> ) JRE: we probably should state that relationship names that include a semicolon need to be quoted. Relationship values are URIs that identify the type of link. If the relationship is a relative URI, its base URI MUST be considered to be "http://www.iana.org/assignments/link-relations.html#", and the value MUST be present in the link relation registry. JRE: why a new base URI? What's wrong with "http://www.iana.org/assignments/relation/" (<http://greenbytes.de/tech/webdav/rfc4287.html#rfc.section.4.2.7.2>)? 6.1. Normative References JRE: I think RFC2434 and RFC3864 could be classified as informative. to map the profiled link relations to URIs. For example, in HTML: <html> <head profile="http://example.com/profile1/"> <link rel="foo" href="/foo"> </head> [...] could be represented as a header like this; Link: </foo>; rel="http://example.com/profile1/foo" JRE: do we need to talk about profile URIs where concatenation does not work well, such as "http://www.w3.org/2003/g/data-view"? HTML defines link relation values as case-insensitive, while the Link header's syntax does not. Therefore, it is important to case- normalise relation values in HTML before comparing or converting them to Link headers. JRE: impact on registration procedure? Avoid names that only differ in case? Atom conveys links in the atom:link element. When serialising an atom:link into a Link header, it is necessary to convert any IRIs to URIs, since HTTP headers cannot directly contain UTF-8. JRE: that's a bit misleading. For instance, a IRI that contains non-ASCII Latin-1 characters could be put into an HTTP header, we just don't want to allow that and chose URIs as format. Additionally, since the base URI for link relations in Link headers is fixed, extension links (i.e,. those not in the registry) MUST be serialised as absolute URIs. JRE: s/serialisized/represented/? BR, Julian
Received on Friday, 14 March 2008 13:12:25 UTC