1.1 Document Conventions

The formal grammar for locators is given using a simple Extended Backus-Naur Form (EBNF) location, as described in the XML specification.

XLink has adopted the following conventions for non-normative commentary:

Issue (Issue Name):

A recorded issue and something on which the working group is seeking comment

Note:

General comments the editors wish to call to the attention of the reader

1.2 Origin and Goals

Following is a summary of the design principles governing XLink, derived from but not overriding the XLink Requirements Document:

1. XLink must be straightforwardly usable over the Internet.
2. XLink must be usable by a wide variety of link usage domains and classes of linking application software.
3. XLink must support HTML 4.0 linking constructs.

   Issue (HTML Support):

   The XLink WG has not determined to what extent the current draft supports HTML linking constructs, nor to what extent we will in the final document. We welcome feedback on this issue.

4. The XLink expression language must be XML.
5. The XLink design must be formal, concise, and illustrative.
6. XLinks must be human-readable and human-writable.
7. XLinks may reside within or outside the documents in which the participating resources reside.
8. XLink must represent the abstract structure and significance of links.
9. XLink must be feasible to implement regardless of the media used for the presentation of links.
10. XLink must be informed by knowledge of established hypermedia systems and standards.

1.3 Relationship to Existing Standards

Three standards have been especially influential:

• HTML: Defines several SGML element types that represent links.
• HyTime: Defines inline and out-of-line link structures and some semantic features, including traversal control and presentation of objects.
• Text Encoding Initiative Guidelines (TEI P3): Provides structures for creating links, aggregate objects, and link collections out of them.

Many other linking systems have also informed this design, especially Dexter, FRESS, MicroCosm, and InterMedia.

1.4 Terminology

The following basic terms apply in this document.

Issue (infoset):

Shouldn't XLink also be described in terms of the XML Information set? See http://www.w3.org/TR/xml-infoset

[Definition: ] arc

A symbolic representation of traversal behavior in links, especially the direction, context and timing of traversal.

[Definition: ] element tree

A representation of the relevant structure specified by the tags and attributes in an XML document.

[Definition: ] hyperlink

An explicit relationship between two or more data objects or portions of data objects with an eye to presentation and traversal in user interfaces.

[Definition: ] inclusion

The process of traversing a link for the purpose of copying the target element tree into the source element tree.

[Definition: ] inline link

Abstractly, a link which serves as one of its own resources. Concretely, a link where the content of the linking element serves as a participating resource. HTML A, HyTime clink, and TEI XREF are all inline links.

[Definition: ] link

An explicit relationship between two or more data objects or portions of data objects.

[Definition: ] linking element

An element that asserts the existence and describes the characteristics of a link.

[Definition: ] local resource

The content of an inline linking element. Note that the content of the linking element could be explicitly pointed to by means of a regular locator in the same linking element, in which case the resource is considered remote, not local.

[Definition: ] locator

Data, provided as part of a link, which identifies a remote resource.

[Definition: ] multidirectional link

A link whose traversal can be initiated from more than one of its participating resources. Note that being able to "go back" after following a one-directional link does not make the link multidirectional.

[Definition: ] out-of-line link

A link whose content does not serve as one of the link's participating resources . Such links presuppose a notion like extended link groups, which instruct application software where to look for links. Out-of-line links are generally required for supporting multidirectional traversal and for allowing read-only resources to have outgoing links.

[Definition: ] parsed

In the context of link behavior, a parsed link is any link whose content is included into the document where the link originated. The use of the term "parsed" directly refers to the concept in XML of a parsed entity.

[Definition: ] participating resource

A resource that belongs to a link. All resources are potential contributors to a link; participating resources are the actual contributors to a particular link.

[Definition: ] remote resource

Any participating resource of a link that is pointed to with a locator.

[Definition: ] resource

In the abstract sense, an addressable unit of information or service that is participating in a link. Examples include files, images, documents, programs, and query results. Concretely, anything reachable by the use of a locator in some linking element. Note that this term and its definition are taken from the basic specifications governing the World Wide Web, such as IETF RFCs [IETF RFC 2396], [IETF RFC 1738] and [IETF RFC 1808].

[Definition: ] sub-resource

A portion of a resource, pointed to as the precise destination of a link. As one example, a link might specify that an entire document be retrieved and displayed, but that some specific part(s) of it is the specific linked data, to be treated in an application-appropriate manner such as indication by highlighting, scrolling, etc.

[Definition: ] traversal

The action of using a link; that is, of accessing a resource. Traversal may be initiated by a user action (for example, clicking on the displayed content of a linking element) or occur under program control.

Issue (hypertext link versus hyperlink):

Is there a difference between hypertext links and hyperlinks?

4.1 Locator Attributes

The only locator attribute at this time is href. This attribute value must be a URI-reference as defined in [IETF RFC 2396].

4.2 Arc End Attributes

Arcs contain two end attributes, from and to. The from attribute may contain a string containing the content of an ID attribute from the resource being linked from. The purpose of the from attribute is to define where this link is being actuated from.

The to attribute may contain a string containing the content of an ID attribute from the resource being linked to. The purpose of the to attribute is to define where this link traverses to.

The application may use this information in a number of ways, especially in a complex hypertext system, but it is mainly useful in providing context for application behavior. The arc construct is intended to provide an explicit model for the relationship between two locators. The arc construct is the correct construct to associate XLink and non-XLink related traversal information.

Issue (Arcs to IDs or roles):

We changed the suggested method of identifying the from and to values from roles to IDs, making the from and to IDREFs. This disallows the specification of a model of traversal from and to classes of ends, which seems like the too high a level of abstraction (though admittedly this is just a hunch). We wanted to allow arcs to specify locator IDs as their from and to values because this would be giving specific instructions for any two instances of locators.

Issue (Arc containment):

Can an arc exist outside the scope of a link? When arcs to and from were connected with roles, it was required that arcs be contained within the link because the roles of the locators were only valid within the link. But using IDs for arcs means that arcs could be outside the context of a link and it's containing locators.

4.3 Behavior Attributes

There are two attributes associated with behavior: show and actuate. The show attribute defines how the remote resource is to be revealed to the user. It has three options: new, parsed, and replace. The new option indicates that the remote resource should be shown in a new window (or other device context) without replacing the previous content. The parsed option, relating directly to the XML concept of a parsed entity, indicates that the content should be integrated into the document from which the link was actuated. The replace option is the one most commonly seen on the World Wide Web, where the document being linked from is entirely replaced by the object being linked to.

The actuate attribute defines how the link is initiated. It has two options: user and auto. The user option indicates that the link must be initiated by some sort of human-initiated selection, such as clicking on an HTML anchor. The auto option indicates that the link is automatically initiated when the application deems that the user has reached the link. It then follows the behavior set out in the show option.

4.4 Semantic Attributes

There are two attributes associated with semantics, role and title. The role attribute is a generic string used to describe the function of the link's content. For example, a poem might have a link with a role="essay".

The title attribute is designed to provide human-readable text describing the link. It is very useful for those who have text-based applications, whether that be due to a constricted device that cannot display the link's content, or if it's being read by an application to a visually-impaired user, or if it's being used to create a table of links. The title attribute should contain a simple, descriptive string.

5.1 Simple Links

[Definition: ] Simple links can be used for purposes that approximate the functionality of a basic HTML A link, but they can also support a limited amount of additional functionality. Simple links have only one locator and thus, for convenience, combine the functions of a linking element and a locator into a single element. As a result of this combination, the simple linking element offers both a locator attribute and all of the behavior and semantic attributes.

The following are two examples of linking elements, each showing all the possible attributes that can be associated with a simple link. Here is the explicit XLink simple linking element.

<!ELEMENT xlink:simple ANY> <!ATTLIST xlink:simple href CDATA #REQUIRED role CDATA #IMPLIED title CDATA #IMPLIED show (new|parsed|replace) "replace" actuate (user|auto) "user" >

And here is how to make an arbitrary element into a simple link.

<!ELEMENT student ANY> <!ATTLIST student xmlns:xlink CDATA #FIXED "http://www.w3.org/XML/XLink/0.9" xlink:type (simple|extended|locator|arc) #FIXED "simple" xlink:href CDATA #REQUIRED xlink:role CDATA #IMPLIED xlink:title CDATA #IMPLIED xlink:show (new|parsed|replace) "replace" xlink:actuate (user|auto) "user" >

Here is how the first example might look in a document:

<xlink:simple href="students.xml" role="student list" title="Student List" show="replace" actuate="user"> Current List of Students </xlink:simple>

Here is how the second example might look in a document:

<students xmlns:xlink="http://www.w3.org/XML/XLink/0.9" xlink:href="students.xml" xlink:role="student list" xlink:title="Student List" xlink:show="new" xlink:actuate="user"> Current List of Students </students>

Alternately, a simple link could be as terse as this:

<students xmlns:xlink="http://www.w3.org/XML/XLink/0.9" xlink:href="students.xml"> The list of students. </students>

Compare this with a rough equivalent in HTML:

<A href="students.html">The list of students.</A>

There are no constraints on the contents of a simple linking element. In the sample declaration above, it is given a content model of ANY to illustrate that any content model or declared content is acceptable. In a valid document, every element that is significant to XLink must still conform to the constraints expressed in its governing DTD.

Note that it is meaningful to have an out-of-line simple link, although such links are uncommon. They are called "one-ended" and are typically used to associate discrete semantic properties with locations. The properties might be expressed by attributes on the link, the link's element type name, or in some other way, and are not considered full-fledged resources of the link. Most out-of-line links are extended links, as these have a far wider range of uses.

5.2 Extended Links

[Definition: ] An extended link differs from a simple link in that it can connect any number of resources, not just one local resource (optionally) and one remote resource, and in that extended links can be out-of-line

These additional capabilities of extended links are required for:

• Enabling outgoing links in documents that cannot be modified to add an inline link
• Creating links to and from resources in formats with no native support for embedded links (such as most multimedia formats)
• Applying and filtering sets of relevant links on demand
• Enabling other advanced hypermedia capabilities

Application software might be expected to provide traversal among all of a link's participating resources (subject to semantic constraints outside the scope of this specification) and to signal the fact that a given resource or sub-resource participates in one or more links when it is displayed (even though there is no markup at exactly that point to signal it).

A linking element for an extended link contains a series of child elements that serve as locators and arcs. Because an extended link can have more than one remote resource, it separates out linking itself from the mechanisms used to locate each resource (whereas a simple link combines the two).

The xlink:type attribute value for an extended link must be extended, if the link is being instantiated on an arbitrary element. Note that extended links introduce variants of the show and actuate behavior attributes. These attributes, the showdefault and actuatedefault define the same behavior as their counterparts. However, in this case, they are considered to define the default behavior for all the linking elements that they contain.

However, when a linking element within an extended link has a show or actuate attribute of its own, that attribute overrides the defaults set on the extended linking element.

The extended linking element itself retains those attributes relevant to the link as a whole and to any local resource. Following are two sample declarations for an extended link. The first is an example of the explicit XLink extended link:

<!ELEMENT xlink:extended ((xlink:arc | xlink:locator)*)> <!ATTLIST xlink:extended role CDATA #IMPLIED title CDATA #IMPLIED showdefault (new|parsed|replace) #IMPLIED actuatedefault (user|auto) #IMPLIED >

The second is an example of an arbitrary element being used an extended link:

<!ELEMENT students ((xlink:arc | xlink:locator)*)> <!ATTLIST students xmlns:xlink CDATA #FIXED "http://www.w3.org/XML/XLink/0.9" xlink:type (simple|extended|locator|arc) #FIXED "extended" xlink:role CDATA #IMPLIED xlink:title CDATA #IMPLIED xlink:showdefault (new|parsed|replace) #IMPLIED xlink:actuatedefault (user|auto) #IMPLIED >

The following two examples demonstrate how each of the above might appear within a document instance. Note that the content of these examples would be other elements. The first example shows how the link might appear, using an explicit XLink extended link:

<xlink:extended role="student list" title="Student List" showdefault="replace" actuatedefault="user"> ... </xlink:extended>

And the second shows how the link might appear, using an arbitrary element and containing additional content:

<students xmlns:xlink="http://www.w3.org/XML/XLink/0.9" xlink:type="extended" xlink:role="student list" xlink:title="Student List" xlink:showdefault="replace" xlink:actuatedefault="user"> ... </students>

Issue (Order of Arcs and Locators):

Should the order of arcs and locators (arcs before locators versus arcs after locators) be specified?

5.3 Locator Elements

[Definition: ] A locator is contained within an extended link for the purpose of defining resources that are participating in a link. The remote resource is specified by the href attribute. More than one locator may be associated with a link. Each remote resource can have its own semantics in relation to the link as a whole and these are specified in the locator element. These are the title and role attributes.

<!ELEMENT xlink:locator ANY>
<!ATTLIST xlink:locator
	id			ID				#REQUIRED
	href			CDATA				#REQUIRED
	role			CDATA				#IMPLIED
	title			CDATA				#IMPLIED >

<!ELEMENT student ANY>
<!ATTLIST student
	xmlns:xlink		CDATA				#FIXED "http://www.w3.org/XML/XLink/0.9"
	xlink:type		(locator)			#FIXED "locator"
	id			ID				#REQUIRED
	xlink:href		CDATA				#REQUIRED
	xlink:role		CDATA				#IMPLIED
	xlink:title		CDATA				#IMPLIED >

<xlink:locator href="/essays/student1/FirstEssay" role="essay"
	title="homework assignment 1"/>

<student xmlns:xlink="http://www.w3.org/XML/XLink/0.9"
	xlink:href="FirstEssay">The First Essay.</foo>

Note:

Locators have a subset of the attributes of simple links. It is not feasible to combine the locator and simple link elements together because the semantics of linking versus locating are substantially different.

Note:

Locators are conceptually similar to programming handles. They provide a needed layer of abstraction for associating metadata with the resource.

Issue (locator container or content):

How does a locator specify the container versus specifying the content of the container?

Issue (HTML access key):

Lack of an equivalent in XLink to the HTML 4.0 ACCESSKEY attribute (on anchor, area). Further consideration should also be given to the use of TITLE and/or link content to provide text which can be identified by speech recognition software as a label for the link, thereby allowing it to be activated via a spoken command.

5.4 Arc Elements

[Definition: ] An arc is contained within an extended link for the purpose of defining traversal behavior. More than one arc may be associated with a link. Otherwise, arc elements function exactly as the arc attributes might lead one to expect.

Here is the explicit XLink arc element.

<!ELEMENT xlink:arc ANY> <!ATTLIST xlink:arc from IDREF #REQUIRED to IDREF #REQUIRED >

And here is how to make an arbitrary element into an arc.

<!ELEMENT genericarc ANY> <!ATTLIST genericarc xmlns:xlink CDATA #FIXED "http://www.w3.org/XML/XLink/0.9" xlink:type (arc) ` #FIXED "arc" xlink:from IDREF #REQUIRED xlink:to IDREF #REQUIRED show (new|parsed|replace) "replace" actuate (user|auto) "user" >

Here is how the first example might look in a document:

<xlink:arc from="student" to="essay" show="parsed" actuate="auto"/>

Here is how the second example might look in a document:

<genericarc xmlns:xlink="http://www.w3.org/XML/XLink/0.9" xlink:type="arc" xlink:from="student" xlink:to="teacher" show="new" actuate="user"/>

Default arc instructions are on the linking element - the show/actuate values inherit from the values in the parent. The alternative is to have a special "default arc" element but that prevents us from setting partial defaults, like "from any id" or "to any id". The "to any id" default could be done on locator elements instead, but this would conflate arcs and ends. The show and actuate attributes on the arc are optional. We proposed that there be cascading defaults, starting with defaults set by the specification, overridden by defaults (if any) set on the xlink element, overridden by values (if any) set on individual arc elements.

Issue (Empty Arc Content):

Arcs are empty for simplicity. What purpose could content serve? Or should we allow content and decide that an XLink processor simply ignores it?

Note:

We proposed not putting role and title attributes on arc elements because we felt that the role of any arc is the concatenation of the two end roles, and if you're going to display title text for an end, there's no sensible place to display the title text for an arc in addition. In cases where the relationship between two elements has a role, then this mechanism does not hold. For example, a family may have many members, and the arc between a one person and another person could have a daughter role, whereas the arc in the opposite direction could have a mother role.

Note:

Every pair of locators has 4 implicit arcs: From each locator to itself, and the two arcs going opposite directions between the locators. An XLink processor is required to make the arcs explicit after processing the link. This enables an application to perform behavior based upon either implicit or explicit arcs. There is no need to specify an arc if the default arcs between the locators is sufficient for traversal.

5.5 Extended Link Groups

[Definition: ] An extended link group specifies an XML document that may contain extended links and additional extended link groups. Hyperlinked documents may be processed in groups rather than one at a time. If it is desired to highlight resources to advertise that traversal can be initiated, and if at the same time out-of-line links are being used, it may be an absolute requirement to read other documents to find these links and discover where the resources are.

In these cases, an extended link group element, a special kind of extended link, may be used to store a list of links to other documents that together constitute an interlinked group.[Definition: ] Each such document is identified by means of an extended link document element, a special kind of locator element.

Following are sample declarations for extended link group and extended link document elements, showing all the possible XLink-related attributes they may have. The xlink:type attribute value for an extended link group element must be group, and the value for an extended link document element must be document.

<!ELEMENT xlink:group (xlink:document*)> <!ATTLIST xlink:group steps CDATA #IMPLIED > <!ELEMENT xlink:document EMPTY> <!ATTLIST xlink:document href CDATA #REQUIRED >

<xlink:group steps="2"> <xlink:document href="http://www.w3.org/W3CHubDoc.xml"/> </xlink:group>

The steps attribute may be used by an author to help deal with the situation where an extended link group directs application software to locate another document, which proves to contain an extended link group of its own. There is a potential for infinite regress, and yet there are situations where processing several levels of extended link groups is useful. The steps attribute should have a numeric value that serves as a hint from the author to any link processor as to how many steps of extended link group processing should be undertaken. It does not have any normative effect.

For example, should a group of documents be organized with a single "hub" document containing all the out-of-line links, it might make sense for each non-hub document to contain an extended link group containing only one reference to the hub document. In this case, the best value for steps would be 2.

Issue (ELG Steps):

dorchard. Steps = 2 seems wrong to me. I would have though = 1. Steps is the depth of the graph of documents to retrieve from the current position, so 1 would cause retrieval of the hub document.

Note:

There are many possibilities for specifying the graph of the documents to be retrieved - HyTime uses the term Bounded Object Set. The depth of the tree is the simplest and easiest for authors and processors.

In another example, extended links are grouped following a multi-level hierarchy, similar to the Domain Name Service. There is an extended link group for each level of the hierarchy, where the extended link group contains extended links relevant to the current level, extended link groups referencing each sub-level, and an extended link group referencing the parent extended link group.

6.1 Processing Model

There are three levels of XLink processing.

1. Well-formed: All XLinks are well-formed XML 1.0 + namespaces compliant and can be parsed by a parser that supports XML 1.0 and namespaces
2. Valid: All XLinks are valid XML XML 1.0 + namespaces compliant and can be validated by a parser that supports XML 1.0 and namespaces
3. Processing: XLink specifies that a parser must expand all implicit arcs in a link into a node. The processing of an XLink requires an application or an XSL transform and XSL compliant parser. Arc expansion is required for XLink conformant Processor.

Issue (Extended Link Validation):

We can't validate XLinks with a mandatory locator and optional arcs + locators. What does the DTD look like?

Issue (Link Processing):

A specific subset of the XSL Integration overlap issue, is should a set of required XLink processing constructs be added to XLink to support inclusion/embedding prior to styling?

Note:

There is no processing model for Extended Link Groups. There is no specification for what a processor is to do when it finds extended links in an extended link group document. A sample list of situations with unspecified results follows:

1. Duplicate IDs between in-line versus the out-of-line links referenced in the extended link group.
2. Retrieval process, especially errors and verbose documents.
3. Circular and self references in the extended link groups.

6.2 Conformance

An element conforms to XLink if:

1. The element has an xlink:type attribute whose value is one of the attribute values prescribed by this specification, and
2. the element and all of its attributes and content adhere to the syntactic requirements imposed by the chosen xlink:type attribute value, as prescribed in this specification.

Note that conformance is assessed at the level of individual elements, rather than whole XML documents, because XLink and non-XLink linking mechanisms may be used side by side in any one document. Additionally, an XLink element may have attributes that are not XLink attributes.

An application conforms to XLink if it interprets XLink-conforming elements according to all required semantics prescribed by this specification and for any optional semantics it chooses to support, supports them in the way prescribed. A user agent application should allow their users to exercise control over document refresh/redirection, especially arcs with actuate="auto".

The required (normative) constructs for a conforming application are simple links, locators, arcs, extended links.