]>

mnot@mnot.net http://www.mnot.net/

This document specifies relation types for Web links, and defines a registry for them. It also defines how to send such links in HTTP headers with the Link header-field.

A means of indicating the relationships between resources on the Web, as well as indicating the type of those relationships, has been available for some time in HTML , and more recently in Atom . These mechanisms, although conceptually similar, are separately specified. However, links between resources need not be format-specific; it can be useful to have typed links that are independent of their serialisation, especially when a resource has representations in multiple formats. To this end, this document defines a framework for typed links that isn't specific to a particular serialisation. It does so by re-defining the link relation registry established by Atom to have a broader scope, and adding to it the relations that are defined by HTML. Furthermore, an HTTP header-field for conveying typed links was defined in , but removed from , due to a lack of implementation experience. Since then, it has been implemented in some User-Agents (e.g., for stylesheets), and several additional use cases have surfaced. Because it was removed, the status of the Link header is unclear, leading some to consider minting new application-specific HTTP headers instead of reusing it. This document addresses this by re-specifying the Link header as one such serialisation, with updated but backwards-compatible syntax. [[ Feedback is welcome on the ietf-http-wg@w3.org mailing list, although this is NOT a work item of the HTTPBIS WG. ]]

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14, , as scoped to those conformance targets. This document uses the Augmented Backus-Naur Form (ABNF) notation of , and explicitly includes the following rules from it: quoted-string, token, SP (space). Additionally, the following rules are included from : URI and URI-Reference, and from : type-name.

In this specification, a link is a typed connection between two resources that are identified by IRIs , and is comprised of: A context IRI, and a link relation type (), and a target IRI, and optionally, target attributes. A link can be viewed as a statement of the form "{context IRI} has a {relation type} resource at {target IRI}, which has {target attributes}." Note that in the common case, the context IRI will also be a URI , because common protocols (such as HTTP) do not support dereferencing IRIs. Likewise, the target IRI will be converted to a URI (see , Section 3.1) in serialisations that do not support IRIs (e.g., the Link header). This specification does not place restrictions on the cardinality of links; there can be multiple links from and to a particular IRI, and multiple links of different types between two given IRIs. Likewise, the relative ordering of links in any particular serialisation, or between serialisations (e.g., the Link header and in-content links) is not specified or significant in this specification; applications that wish to consider ordering significant MAY do so. Target attributes are a set of key/value pairs that describe the link or its target; for example, a media type hint. This specification does not attempt to coordinate their names or use, but does provide common target attributes for use in the Link HTTP header. Finally, this specification does not define a general syntax for expressing links, nor mandate a specific context for any given link; it is expected that serialisations of links will specify both aspects. One such serialisation is communication of links through HTTP headers, specified in .

A link relation type identifies the semantics of a link. For example, a link with the relation type "copyright" indicates that the resource identified by the target IRI is a statement of the copyright terms applying to the current context IRI. Relation types are not to be confused with media types ; they do not identify the format of the representation that results when the link is dereferenced. Rather, they only describe how the current context is related to another resource. As such, relation types are not format-specific, and MUST NOT specify a particular format or media type that they are to be used with. Likewise, the context IRI for a given link is usually determined by the serialisation of the link (e.g., the Link header, a HTML document, etc.); a relation type SHOULD NOT specify the context IRI. Relation types SHOULD NOT infer any additional semantics based upon the presence or absence of another link relation, or its own cardinality of occurrence. An exception to this is the combination of the "alternate" and "stylesheet" registered relation types, which have special meaning in HTML4 for historical reasons. Consuming implementations SHOULD ignore relation types that they do not understand or have no need to process. There are two kinds of relation types; registered and extension.

Commonly-used relation types with a clear meaning that are shared across applications can be registered as tokens for convenience and to promote reuse. For example, "self" and "alternate" are registered relation types, because they are broadly useful. This draft establishes an IANA registry of such relation types; see . Registered relation types MUST conform to the token rule, and SHOULD conform to the sgml-name rule for compatibility with deployed implementations;

sgml-name = ALPHA *( ALPHA | DIGIT | "." | "-" )

Names that differ only in case from existing entries (e.g., "Foo" and "foo") MUST NOT be registered. Registered relation types MUST be compared in a case-insensitive fashion. Although registered relation types are specified as tokens, applications wishing to internally refer to one using a URI MAY do so by considering it relative to the base URI "http://www.iana.org/assignments/relation/". However, the URI form of a registered relation type SHOULD NOT be serialised when an application specifies the use of a relation type, because a consuming implementation may not recognise it.

Applications that don't merit a registered relation type may use an extension relation type, which is a URI that uniquely identifies the relation type. Although the URI MAY point to a resource that contains a definition of the semantics of the relation type, clients SHOULD NOT access that resource to avoid overburdening its server. When extension relation types are compared, they MUST be compared as URIs in a case-sensitive fashion, character-by-character. Note that while extension relation types are required to be URIs, but a serialisation of links MAY specify that they are expressed in another form, as long as they can be converted to URIs.

The Link entity-header field provides a means for serialising one or more links in HTTP headers. It is semantically equivalent to the <LINK> element in HTML, as well as the atom:link feed-level element in Atom .

" *( ";" link-param ) link-param = ( ( "rel" "=" relation-types ) | ( "rev" "=" relation-types ) | ( "type" "=" type-name ) | ( "title" "=" quoted-string ) | ( "title*" "=" enc2231-string ) | ( "anchor" "=" <"> URI-Reference <"> ) | ( link-extension ) ) link-extension = token [ "=" ( token | quoted-string ) ] enc2231-string = relation-types = relation-type | <"> relation-type *( SP relation-type ) <"> relation-type = reg-relation-type | ext-relation-type reg-relation-type = token ext-relation-type = URI ]]>

Each link-value conveys one target IRI as a URI-Reference (after conversion to one, if necessary) inside angle brackets ("<>"). If the URI-Reference is relative, it MUST be resolved as per , Section 5. Note that any base IRI from the body's content is not applied. By default, the context of a link conveyed in the Link header field is the IRI of the requested resource. When present, the anchor parameter overrides this with another URI, such as a fragment of this resource, or a third resource (i.e., when the anchor value is an absolute URI). If the anchor parameter's value is a relative URI, it MUST be resolved as per , Section 5. Note that any base URI from the body's content is not applied. Normally, the relation type of a link is conveyed in the "rel" parameter's value. The "rev" parameter has also been used for this purpose historically by some formats, and is included here for compatibility with those uses, but its use is not encouraged nor defined by this specification. Note that extension relation types are REQUIRED to be absolute URIs in Link headers, and MUST be quoted if they contain a semicolon (";") or comma (","). The "title", "title*" and any link-extension link-params are considered to be the target parameters for the link. The "title" parameter is used to label the destination of a link such that it can be used as a human-readable identifier (e.g. a menu entry). Alternately, the "title*" parameter MAY be used encode this label in a different character set, and/or contain language information as per . When using the enc2231-string syntax, producers MUST NOT use a charset value other than 'ISO-8859-1' or 'UTF-8'.

NOTE: Non-ASCII characters used in prose for examples are encoded using the format "Backslash-U with Delimiters", defined in Section 5.1 of . For example:

; rel="previous"; title="previous chapter" ]]>

indicates that "chapter2" is previous to this resource in a logical navigation path. The example below shows an instance of the Link header encoding multiple links, and also the use of RFC 2231 encoding to encode both non-ASCII characters and language information.

; rel="previous"; title*=UTF-8'de'letztes%20Kapitel", ; rel="next"; title*=UTF-8'de'n%c3%a4chstes%20Kapitel" ]]>

Here, the second link has a title encoded in UTF-8, uses the German language ("de"), and contains the Unicode code point \u'00E4' ("LATIN SMALL LETTER A WITH DIAERESIS"). Note that link-values may convey multiple links between the same target and context IRIs; for example:

; rel=index; rel="start http://example.net/relation/other" ]]>

Here, the link to "http://example.org/" has the registered relation types "index" and "start", and the extension relation type "http://example.net/relation/other".

This specification updates the Message Header Registry entry for "Link" in HTTP to refer to this document.

Header field: Link Applicable protocol: http Status: standard Author/change controller: IETF (iesg@ietf.org) Internet Engineering Task Force Specification document(s): [ this document ]

This specification establishes the Link Relation Type Registry, located at , and updates Atom to refer to it in place of the "Registry of Link Relations". The requirements for registered relation types are described in . Relation types may be registered on the advice of a Designated Expert (appointed by the IESG or their delegate), with a Specification Required (using terminology from ). Registration requests consist of the completed registration template below, typically published in an RFC or Open Standard (in the sense described by , Section 7). However, to allow for the allocation of values prior to publication, the Designated Expert may approve registration once they are satisfied that an RFC (or other Open Standard) will be published. The registration template is: Relation Name: Description: Reference: Upon receiving a registration request (usually via IANA), the Designated Expert should request review and comment from the apps-discuss@ietf.org mailing list (or a successor designated by the APPS Area Directors). Before a period of 30 days has passed, the Designated Expert will either approve or deny the registration request, communicating this decision both to the review list and to IANA. Denials should include an explanation and, if applicable, suggestions as to how to make the request successful. The Link Relation Type registry's initial contents are: Relation Name: alternate Description: Designates a substitute for the link's context. Reference: Relation Name: appendix Description: Refers to an appendix. Reference: Relation Name: bookmark Description: Refers to a bookmark or entry point. Reference: Relation Name: chapter Description: Refers to a chapter in a collection of resources. Reference: Relation Name: contents Description: Refers to a table of contents. Reference: Relation Name: copyright Description: Refers to a copyright statement that applies to the link's context. Reference: Relation Name: current Description: Refers to a resource containing the most recent item(s) in a collection of resources. Reference: Relation Name: describedby Description: Refers to a resource providing information about the link's context. Documentation: Relation Name: edit Description: Refers to a resource that can be used to edit the link's context. Reference: Relation Name: edit-media Description: Refers to a resource that can be used to edit media associated with the link's context. Reference: Relation Name: enclosure Description: Identifies a related resource that is potentially large and might require special handling. Reference: Relation Name: first Description: An IRI that refers to the furthest preceding resource in a series of resources. Reference: [this document] Relation Name: glossary Description: Refers to a glossary of terms. Reference: Relation Name: help Description: Refers to a resource offering help (more information, links to other sources information, etc.) Reference: Relation Name: index Description: Refers to an index. Reference: Relation Name: last Description: An IRI that refers to the furthest following resource in a series of resources. Reference: [this document] Relation Name: license Description: Refers to a license associated with the link's context. Reference: Relation Name: next Description: Refers to the next resource in a ordered series of resources. Reference: Relation Name: next-archive Description: Refers to the immediately following archive resource. Reference: Relation Name: payment Description: indicates a resource where payment is accepted. Reference: [this document] Relation Name: prev Description: Refers to the previous resource in an ordered series of resources. Synonym for "previous". Reference: Relation Name: previous Description: Refers to the previous resource in an ordered series of resources. Synonym for "prev". Reference: Relation Name: prev-archive Description: Refers to the immediately preceding archive resource. Reference: Relation Name: related Description: Identifies a related resource. Reference: Relation Name: replies Description: Identifies a resource that is a reply to the context of the link. Reference: Relation Name: section Description: Refers to a section in a collection of resources. Reference: Relation Name: self Description: Conveys an identifier for the link's context. Reference: Relation Name: start Description: Refers to the first resource in a collection of resources. Reference: Relation Name: stylesheet Description: Refers to an external style sheet. Reference: Relation Name: subsection Description: Refers to a resource serving as a subsection in a collection of resources. Reference: Relation Name: via Description: Identifies a resource that is the source of the information in the link's context. Reference:

The content of the Link header-field is not secure, private or integrity-guaranteed, and due caution should be exercised when using it. Applications that take advantage of typed links should consider the attack vectors opened by automatically following, trusting, or otherwise using links gathered from HTTP headers. In particular, Link headers that use the "anchor" parameter to associate a link's context with another resource should be treated with due caution.

Target IRIs may need to be converted to URIs in order to express them in serialisations that do not support IRIs. This includes the Link HTTP header. Similarly, the anchor parameter of the Link header does not support IRIs, and therefore IRIs must be converted to URIs before inclusion there. Relation types are defined as URIs, not IRIs, to aid in their comparison. It is not expected that they will be displayed to end users.

&rfc2026; &rfc2119; &rfc2231; &rfc3986; &rfc3987; &rfc2616; &rfc3864; &rfc4288; &rfc5226; &rfc2068; &rfc4287; &rfc4685; &rfc4946; &rfc5005; &rfc5023; &rfc5137; &html;

HTML motivated the original syntax of the Link header, and many of the design decisions in this document are driven by a desire to stay compatible with these uses. In HTML4, the link element can be mapped to links as specified here by using the "href" attribute for the target URI, and "rel" to convey the relation type, as in the Link header. The context of the link is the URI associated with the entire HTML document. HTML4 also has a "rev" parameter for links that allows a link's relation to be reversed. The Link header has a "rev" parameter to allow the expression of these links in HTTP headers, but its use is not encouraged, due to the confusion this mechanism causes as well as conflicting interpretations among HTML versions. All of the link relations defined by HTML4 have been included in the link relation registry, so they can be used without modification. However, extension link relations work differently in HTML4 and the Link header; the former uses a document-wide "profile" URI to scope the relations, while the latter allows the use of full URIs on individual relations. Therefore, when using the profile mechanism in HTML4, it is necessary to map the profiled link relations to URIs when expressed in Link headers. For example, in HTML:

[...] ]]>

could be represented as a header like this;

Link: </bar>; rel="http://example.com/profile1/foo"

Profile authors should note this when creating profile URIs; it may be desirable to use URIs that end in a delimiter (e.g., "/" or "#"), to make extracting the specific relation in use easier. Surveys of existing HTML content have shown that unregistered link relation types that are not URIs are (perhaps inevitably) common. Consuming HTML implementations should not consider such unregistered short links to be errors, but rather relation types with a local scope (i.e., their meaning is specific and perhaps private to that document). HTML4 also defines several attributes on links that are not explicitly defined by the Link header. These attributes can be serialised as link-extensions to maintain fidelity. Finally, the HTML4 specification gives a special meaning when the "alternate" and "stylesheet" relations coincide in the same link. Such links should be serialised in the Link header using a single list of relation-types (e.g., rel="alternate stylesheet") to preserve this relationship.

Atom conveys links in the atom:link element, with the "href" attribute indicating the target IRI and the "rel" attribute containing the relation type. The context of the link is either a feed IRI or an entry ID, depending on where it appears; generally, feed-level links are candidates for transmission as a Link header. When serialising an atom:link into a Link header, it is necessary to convert target IRIs (if used) to URIs. Atom defines extension relation types in terms of IRIs. This specification defines them as URIs, to aid in their comparison. Atom allows registered link relation types to be serialised as absolute URIs, because a base URI is defined for the registry. Such relation types SHOULD be converted to the appropriate registered form (e.g., "http://www.iana.org/assignments/relation/self" to "self") so that they are not mistaken for extension relation types. Furthermore, Atom link relations are compared in a case-sensitive fashion; therefore, registered link relations SHOULD be converted to their registered form (usually, lower case) when serialised in an Atom document. Note also that while the Link header allows multiple relations to be associated with a single link, atom:link does not. In this case, a single link-value may map to several atom:link elements. As with HTML, atom:link defines some attributes that are not explicitly mirrored in the Link header syntax, but they may also be used as link-extensions.

New serialisations of links (as defined by this specification) need to address several issues, including: Specific syntax for each component of the link model described in . What target attributes, if any, are defined by the serialisation. How to determine the context of the link. How to differentiate registered link relations from extension link relations (if the latter are serialised as URIs, this is relatively straightforward).

This specification lifts the idea and definition for the Link header from RFC2068; credit for it belongs entirely to the authors of and contributors to that document. The link relation registrations themselves are sourced from several documents; see the applicable references. The author would like to thank the many people who commented upon, encouraged and gave feedback to this draft, especially including Frank Ellermann, Roy Fielding and Julian Reschke.

[[ to be removed by the RFC editor before publication as an RFC. ]] -05 Clarified how to resolve relative URIs in the 'anchor' parameter. Tweaked language about dereferencing relation type URIs. Separated out examples. Made target-parameters more explicit in the model. Discourage special semantics between different relations, or based upon cardinality. Grandfathered in special semantics of 'alternate stylesheet' for HTML4. Note that extension types can be serialised in ways other than as URIs, as long as they can be converted to URIs. Change default context of a link header to that of the requested resource. Use this document as reference for relations that don't have a formal definition other than the registry entries; avoids circular references. Noted that ordering of links is not significant or defined in this spec, but may be in specific applications. Adjusted uses of 'application' to 'serialisation' where appropriate. Added 'Defining New Link Serialisations' section. Added note about case sensitivity when comparing registered relation types in Atom. -04 Defined context as a resource, rather than a representation. Removed concept of link directionality; relegated to a deprecated Link header extension. Relation types split into registered (non-URI) and extension (URI). Changed wording around finding URIs for registered relation types. Changed target and context URIs to IRIs (but not extension relation types). Add RFC2231 encoding for title parameter, explicit BNF for title*. Add i18n considerations. Specify how to compare relation types. Changed registration procedure to Designated Expert. Softened language around presence of relations in the registry. Added describedby relation. Re-added 'anchor' parameter, along with security consideration for third-party anchors. Softened language around HTML4 attributes that aren't directly accommodated. Various tweaks to abstract, introduction and examples. -03 Inverted focus from Link headers to link relations. Specified was a link relation type is. Based on discussion, re-added 'rev'. Changed IESG Approval to IETF Consensus for relation registrations (i.e., require a document). Updated RFC2434 reference to RFC5226. Registered relations SHOULD conform to sgml-name. Cautioned against confusing relation types with media types. -02 Dropped XLink language. Removed 'made' example. Removed 'rev'. Can still be used as an extension. Added HTML reference to introduction. Required relationship values that have a ; or , to be quoted. Changed base URI for relation values. Noted registry location. Added advisory text about HTML profile URIs. Disallowed registration of relations that only differ in case. Clarified language about IRIs in Atom. Added descriptions for 'first', 'last', and 'payment', referring to current IANA registry entries, as these were sourced from e-mail. Will this cause self-referential implosion? Explicitly updates RFC4287. Added 'type' parameter. Removed unnecessary advice about non-HTML relations in HTML section. -01 Changed syntax of link-relation to one or more URI; dropped Profile. Dropped anchor parameter; can still be an extension. Removed Link-Template header; can be specified by templates spec or elsewhere. Straw-man for link relation registry. -00 Initial draft; normative text lifted from RFC2068.