This document defines methods for exchanging instant messages that contain lightweight text markup. In the context of this document, "lightweight text markup" is to be understood as a combination of minimal structural elements and presentational styles that can easily be rendered on a wide variety of devices without requiring a full rich-text rendering engine such as a web browser. Examples of lightweight text markup include basic text blocks (e.g., paragraphs and blockquotes), structural elements (e.g., emphasis and strength), lists, hyperlinks, image references, and font styles (e.g., sizes and colors).

Note: This specification essentially defines a recommended set of XHTML elements and attributes for use in instant messaging by applying a series of "filters" to the wealth of XHTML features; however, developers of Jabber/XMPP clients mainly need to pay attention to the Summary of Recommendations rather than the complexities of how the recommendations are derived.

In the past, there have existed several incompatible methods within the Jabber community for exchanging instant messages that contain lightweight text markup. The most notable such methods have included derivatives of &w3xhtml; as well as of &rtf;.

Although it is sometimes easier for client developers to implement RTF support (this is especially true on certain Microsoft Windows operating systems), there are several reasons (consistent with the &xep0134;) for the &XSF; to avoid the use of RTF in developing a protocol for lightweight text markup. Specifically:

1. RTF is not a structured vocabulary derived from SGML (as is &w3html;) or, more relevantly, from XML (as is XHTML 1.0).
2. RTF is under the control of the Microsoft Corporation and thus is not an open standard maintained by a recognized standards development organization; therefore the XSF is unable to contribute to or influence its development if necessary, and any protocol the XSF developed using RTF would introduce unwanted dependencies.

Conversely, there are several reasons to prefer XHTML for lightweight text markup:

1. XHTML is a structured format that is defined as an application of &w3xml;, making it especially appropriate for sending over Jabber/XMPP, which is at root a technology for streaming XML (see &xmppcore;).
2. XHTML is an open standard developed by the &W3C;, a recognized standards development organization.

Therefore, this document defines support for lightweight text markup in the form of an XMPP extension that encapsulates content defined by an XHTML 1.0 Integration Set that we label "XHTML-IM". The remainder of this document discusses lightweight text markup in terms of XHTML 1.0 only and does not further consider RTF or other technologies.

HTML was originally designed for authoring and presenting stuctured documents on the World Wide Web, and was subsequently extended to handle more advanced functionality such as image maps and interactive forms. However, the requirements for publishing documents (or developing transactional websites) for presentation by dedicated XHTML user agents on traditional computers or small-screen devices are fundamentally different from the requirements for lightweight text markup of instant messages; for this reason, only a reduced set of XHTML features is needed for XHTML-IM. In particular:

1. IM clients are not XHTML clients: their primary purpose is not to read pre-existing XHTML documents, but to read and generate relatively large numbers of fairly small instant messages.

2. The underlying context for XHTML content in Jabber/XMPP instant messaging is provided not by a full XHTML document, but by an XML stream, and specifically by a message stanza within that stream. Thus the <head/> element and all its children are unnecessary. Only the <body/> element and some of its children are appropriate for use in instant messaging.

3. The XHTML content that is read by one's IM client is normally generated on the fly by one's conversation partner (or, to be precise, by his or her IM client). Thus there is an inherent limit to the sophistication of the XHTML markup involved. Even in normal XHTML documents, fairly basic structural and rendering elements such as definition lists, abbreviations, addresses, and computer input handling (e.g., <kbd/> and <var/>) are relatively rare. There is little or no foreseeable need for such elements within the context of instant messaging.

4. The foregoing is doubly true of more advanced markup such as tables, frames, and forms (however, there exists an XMPP extension that provides an instant messaging equivalent of the latter, as defined in &xep0004;).

5. Although ad-hoc styles are useful for messaging (by means of the 'style' attribute), full support for &w3css; (defined by the <style/> element or a standalone .css file, and implemented via the 'class' attribute) would be overkill since many CSS1 properties (e.g., box, classification, and text properties) were developed especially for sophisticated page layout.

6. Background images, audio, animated text, layers, applets, scripts, and other multimedia content types are unnecessary, especially given the existence of XMPP extensions such as &xep0096;, &xep0166;, and &xep0234;.

7. Content transformations such as those defined by &w3xslt; must not be necessary in order for an instant messaging application to present lightweight text markup to an end user.

As explained below, some of these requirements are addressed by the definition of the XHTML-IM Integration Set itself, while others are addressed by a recommended "profile" for that Integration Set in the context of instant messaging applications.

This document defines an adaptation of XHTML 1.0 (specifically, an XHTML 1.0 Integration Set) that makes it possible to provide lightweight text markup of instant messages (mainly for Jabber/XMPP instant messages, although the Integration Set defined herein could be used by other protocols). This pattern is familiar from email, wherein the HTML-formatted version of the message supplements but does not supersede the text-only version of the message. The XHTML is merely an alternative version of the message body or bodies, and the semantic meaning is to be derived from the textual message body or bodies rather than the XHTML version.

In Jabber/XMPP communications, the meaning (as opposed to markup) of the message MUST always be represented as best as possible in the normal <body/> child element or elements of the &MESSAGE; stanza qualified by the 'jabber:client' (or 'jabber:server') namespace. Lightweight text markup is then provided within an <html/> element qualified by the 'http://jabber.org/protocol/xhtml-im' namespace. It might have been better to use an element name other than <html/> for the wrapper element; however, changing it would not be backwards-compatible with the older protocol and existing implementations. However, this <html/> element is used solely as a "wrapper" for the XHTML content itself, which content is encapsulated via one or more <body/> elements qualified by the 'http://www.w3.org/1999/xhtml' namespace, along with appropriate child elements thereof.

The following example illustrates this approach.

Technically speaking, there are three aspects to the approach taken herein:

1. Definition of the <html/> "wrapper" element, which functions as an XMPP extension within XMPP <message/> stanzas.
2. Definition of the XHTML-IM Integration Set itself in terms of supported XHTML 1.0 modules, using the concepts defined in &w3xhtmlmod;.
3. A recommended "profile" regarding the specific XHTML 1.0 elements and attributes to be supported from each XHTML 1.0 module.

These three aspects are defined in the three document sections that follow.

The root element for including XHTML content within XMPP stanzas is <html/>. This element is qualified by the 'http://jabber.org/protocol/xhtml-im' namespace. From the perspective of XMPP, the wrapper element functions as an XMPP extension element; from the perspective of XHTML, it functions as a wrapper for XHTML 1.0 content qualified by the 'http://www.w3.org/1999/xhtml' namespace. Such XHTML content MUST be contained in one or more <body/> elements qualified by the 'http://www.w3.org/1999/xhtml' namespace and MUST conform to the XHTML-IM Integration Set defined in the following section. If more than one <body/> element is included in the <html/> wrapper element, each <body/> element MUST possess an 'xml:lang' attribute with a distinct value, where the value of that attribute MUST adhere to the rules defined in &rfc4646;. A formal definition of the <html/> element is provided in the XHTML-IM Wrapper Schema.

The XHTML <body/> element is not to be confused with the XMPP <body/> element, which is a child of a &MESSAGE; stanza and is qualified by the 'jabber:client' or 'jabber:server' namespace as described in &xmppim;. The <html/> wrapper element is intended for inclusion only as a direct child element of the XMPP &MESSAGE; stanza and only in order to specify a marked-up version of the message &BODY; element or elements, but MAY be included elsewhere in accordance with the "extended namespace" rules defined in the XMPP IM specification.

Until and unless (1) additional integration sets are defined and (2) mechanisms are specified for discovering or negotiating which integration sets are supported, the XHTML markup contained within the <html/> wrapper element:

1. MUST NOT include elements and attributes that are not part of the XHTML-IM Integration Set defined in Section 6 of this document; any such elements and attributes MUST be ignored if received.
2. SHOULD NOT include elements and attributes that are not part of the recommended profile of the XHTML-IM Integration Set defined in Section 7 of this document; any such elements and attributes SHOULD be ignored if received.

Note: In the foregoing restrictions, the meaning of "ignore" is defined by the conformance requirements of Modularization of XHTML, as summarized in the User Agent Conformance section of this document).

This section defines an XHTML 1.0 Integration Set for use in the context of instant messaging. Given its intended usage, we label it "XHTML-IM".

Modularization of XHTML provides the ability to formally define subsets of XHTML 1.0 via the concept of "modularization" (which may be familiar from &w3xhtmlbasic;). Many of the defined modules are not necessary or useful in the context of instant messaging, and in the context of Jabber/XMPP instant messaging specifically some modules have been superseded by well-defined XMPP extensions. This document specifies that XHTML-IM shall be based on the following XHTML 1.0 modules:

• Core Modules
  • Structure Module
  • Text Module
  • Hypertext Module
  • List Module
• Image Module
• Style Attribute Module

Modularization of XHTML defines many additional modules, such as Table Modules, Form Modules, Object Modules, and Frame Modules. None of these modules is part of the XHTML-IM Integration Set. If support for such modules is desired, it MUST be defined in a separate and distinct integration set.

Even within the restricted set of modules specified as defining the XHTML-IM Integration Set (see preceding section), some elements and attributes are inappropriate or unnecessary for the purpose of instant messaging; although such elements and attributes MAY be included in accordance with the XHTML-IM Integration Set, further recommended restrictions regarding which elements and attributes to include in XHTML content are specified below.

The following rules apply to the generation and processing of XHTML content by Jabber clients or other XMPP entities.

1. XHTML-IM content is designed to provide a formatted version of the XML character data provided in the &BODY; of an XMPP &MESSAGE; stanza; if such content is included in an XMPP message, the <html/> element MUST be a direct child of the &MESSAGE; stanza and the XHTML-IM content MUST be understood as a formatted version of the message body. XHTML-IM content MAY be included within XMPP &IQ; stanzas (or children thereof), but any such usage is undefined. In order to preserve bandwidth, XHTML-IM content SHOULD NOT be included within XMPP &PRESENCE; stanzas; however, if it is so included, the <html/> element MUST be a direct child of the &PRESENCE; stanza and the XHTML-IM content MUST be understood as a formatted version of the XML character data provided in the &STATUS; element.

2. The sending client MUST ensure that, if XHTML content is sent, its meaning is the same as that of the plaintext version, and that the two versions differ only in markup rather than meaning.

3. XHTML-IM is a reduced set of XHTML 1.0 and thus also of XML 1.0. Therefore all opening tags MUST be completed by inclusion of an appropriate closing tag.

4. XMPP Core specifies that an XMPP &MESSAGE; MAY contain more than one <body/> child as long as each <body/> possesses an 'xml:lang' attribute with a distinct value. In order to ensure correct internationalization, if an XMPP &MESSAGE; stanza contains more than one <body/> child and is also sent as XHTML-IM, the <html/> element SHOULD also contain more than one <body/> child, with one such element for each <body/> child of the &MESSAGE; stanza (distinguished by an appropriate 'xml:lang' attribute).

5. Section 11.1 of XMPP Core stipulates that character entities other than the five general entities defined in Section 4.6 of the XML specification (i.e., &lt;, &gt;, &amp;, &apos;, and &quot;) MUST NOT be sent over an XML stream. Therefore implementations of XHTML-IM MUST NOT include predefined XHTML 1.0 entities such as &nbsp; -- instead, implementations MUST use the equivalent character references as specified in Section 4.1 of the XML specification (even in non-obvious places such as URIs that are included in the 'href' attribute).

6. For elements and attributes qualified by the 'http://www.w3.org/1999/xhtml' namespace, user agent conformance is guided by the requirements defined in Modularization of XHTML; for details, refer to the User Agent Conformance section of this document.

7. The use of structural elements is NOT RECOMMENDED where presentational styles are desired, which is why very few structural elements are specified herein. Implementations SHOULD use appropriate 'style' attributes (e.g., <span style='font-weight: bold'>this is bold</span> and <p style='margin-left: 5%'>this is indented</p>) rather than XHTML structural elements (e.g., <strong/> and <blockquote/>) wherever possible.

8. Nesting of block structural elements (<p/>) and list elements (<dl/>, <ol/>, <ul/>) is NOT RECOMMENDED, except within <div/> elements.

9. It is RECOMMENDED for implementations to replace line breaks with the <br/> element and to replace significant whitepace with the appropriate number of non-breaking spaces (via the NO-BREAK SPACE character or its equivalent), where "significant whitespace" means whitespace that makes some material difference (e.g., one or more spaces at the beginning of a line or more than one space anywhere else within a line), not "normal" whitespace separating words or punctuation.

10. When rendering XHTML-IM content, a user agent SHOULD NOT render as a hyperlink text that is not structured via the <a/> element from the Hypertext Module; therefore if the sender wishes text to be linked, the sending user agent MUST represent the text using the <a/> element and appropriate attributes.

The following examples provide an insight into the inclusion of XHTML content in XMPP &MESSAGE; stanzas but are by no means exhaustive or definitive.

(Note: The examples might not render correctly in all web browsers, since not all web browsers comply fully with the XHTML 1.0 and CSS1 standards. Markup in the examples may include line breaks for readability. Example renderings are shown with a colored background to set them off from the rest of the text.)

This could be rendered as follows:

This could be rendered as follows:

This could be rendered as follows:

Note the large size of the image. Including the 'height' and 'width' attributes is therefore quite friendly, since it gives the receiving application hints as to whether the image is too large to fit into the current interface (naturally, these are hints only and cannot necessarily be relied upon in determining the size of the image).

Rendering the 'alt' value rather than the image would yield something like the following:

This could be rendered as follows: