MAY"> MUST"> MUST NOT"> OPTIONAL"> RECOMMENDED"> REQUIRED"> SHALL"> SHALL NOT"> SHOULD"> SHOULD NOT"> ]> greenbytes GmbH

Hafenweg 16 MuensterNW48155 Germany julian.reschke@greenbytes.de http://greenbytes.de/tech/webdav/

This document defines the 'XML2RFC' version 2 vocabulary; an XML-based language used for writing RFCs and Internet-Drafts. Discussion of this draft takes place on the XML2RFC mailing list (xml2rfc@ietf.org), which has its home page at .

This document describes version 2 ('v2') of the 'XML2RFC' vocabulary; an XML-based language ('Extensible Markup Language', ) used for writing RFCs () and Internet-Drafts (). It obsoletes the original version ("v1") , which contained the original language definition, and which was subsequently extended ("v2"). Furthermore, it discusses potential extensions in a future revision ("v3"). Note that the vocabulary contains certain constructs that might not be used when generating the final text; however, they can provide useful data for other uses (such index generation, populating a keyword database, or syntax checks).

The XML vocabulary here is defined in prose, based on the Relax NG schema () contained in (specified in Relax NG Compact Notation, "RNC"). Note that the schema can be used for automated validity checks, but certain constraints are only described in prose (example: the conditionally required presence of the "abbrev" attribute).

The sections below describe all elements and their attributes. Note that attributes not labeled "mandatory" are optional.

Contains the abstract of the document. The abstract ought to be self-contained and thus should not contain references or unexpanded abbreviations. See for more information. This element appears as child element of: <front> (). Content model: One or more <t> elements ()

Provides address information for the author. This element appears as child element of: <author> (). Content model: In this order: One optional <postal> element () One optional <phone> element () One optional <facsimile> element () One optional <email> element () One optional <uri> element ()

Provides additional prose augmenting a bibliographical reference.

For instance: <annotation> Latest version available at <eref target='http://www.w3.org/TR/xml'/>. </annotation> ...will generate the text used in the reference for .

This element appears as child element of: <reference> (). Content model: In any order: Text <xref> elements () <eref> elements () <iref> elements () <cref> elements () <spanx> elements ()

Provides information about the IETF area to which this document relates (currently not used when generating documents). The value ought to be either the fullname or the abbreviation of one of the IETF areas as listed on . The list at the time that this document is being published is: "Applications", "app", "General", "gen", "Internet", "int", "Operations and Management", "ops", "Real-time Applications and Infrastructure", "rai", "Routing", "rtg", "Security", "sec", "Transport", "tsv". This element appears as child element of: <front> (). Content model: only text content.

This element allows the inclusion of "artwork" into the document. <artwork> is the only element in the vocabulary that provides full control of horizontal whitespace and line breaks, and thus is used for a variety of things, such as: diagrams ("line art"), source code, formal languages (such as ABNF or the RNC notation used in this document), complex tables, or protocol unit diagrams. Alternatively, the "src" attribute allows referencing an external graphics file, such as a bitmap or a vector drawing, using a URI. In this case, the textual content acts as fallback for output formats that do not support graphics, and thus ought to contain either a "line art" variant of the graphics, or otherwise prose that describes the included image in sufficient detail. Note that RFCs occasionally are published with enhanced diagrams; a recent example is . This element appears as child element of: <figure> (). Content model: Text

Controls whether the artwork appears left (default), centered, or right. Allowed values: "left" (default) "center" "right"

Alternative text description of the artwork (not just the caption).

The suggested height of the graphics included using the "src" attribute. This attribute is format-dependent and ought to be avoided. When generating HTML output, current implementations copy the attribute "as is". For other output formats it is usually ignored.

A filename suitable for the contents (such as for extraction to a local file). This attribute generally isn't used for document generation, but it can be helpful for other kinds of tools (such as automated syntax checkers which work by extracting the source code).

The URI of a graphics file. Note that this can be a "data" URI () as well, in which case the graphics file is wholly part of the XML file.

Specifies the type of the artwork. The value either is a well-known keyword (such as "abnf"), or an Internet Media Type (see ). How it is used depends on context and application. For instance, a formatter can attempt to syntax-highlight code in certain known languages.

The suggested width of the graphics included using the "src" attribute. This attribute is format-dependent and ought to be avoided. When generating HTML output, current implementations copy the attribute "as is". For other output formats it is usually ignored.

Determines whitespace handling. "preserve" is both the default value and the only meaningful setting anyway (because that's what the <artwork> element is for). See also . Allowed values: "default" "preserve" (default)

Provides information about a document's author. This is used both for the document itself (at the beginning of the document) and for referenced documents (inside of <reference>). The <author> elements contained within the document's <front> element are used to fill the boilerplate, and also to generate the "Author's Address" section (see ). Note that an "author" can also be just an organization (by not specifying any of the name attributes, but adding the <organization> child element). Furthermore, the "role" attribute can be used to mark an author as "editor". This is reflected both on the front page and in bibliographical references. Note that this specification does not define a precise meaning for the term "editor". See Section "Authors vs. Contributors" of for more information. This element appears as child element of: <front> (). Content model: In this order: One optional <organization> element () One optional <address> element ()

The full name (used in the automatically generated "Author's Address" section).

Author initials (used on the front page and in references). Initials should be provided as a whitespace separated list of pairs of a letter and a dot.

Specifies the role the author had in creating the document. Allowed values: "editor"

The author's surname.

Contains the "back" part of the document: the references and appendices. In <back>, <section> elements indicate appendices. This element appears as child element of: <rfc> (). Content model: In this order: Optional <references> elements () Optional <section> elements ()

Provides the content of a cell in a table. This element appears as child element of: <texttable> (). Content model: In any order: Text <xref> elements () <eref> elements () <iref> elements () <cref> elements () <spanx> elements ()

Gives the city name in a postal address. This element appears as child element of: <postal> (). Content model: only text content.

Gives the postal region code. This element appears as child element of: <postal> (). Content model: only text content.

Gives the country in a postal address. This element appears as child element of: <postal> (). Content model: only text content.

Represents a comment. Comments can be used in a document while it is work-in-progress. They usually appear either inline and visually highlighted, at the end of the document (depending on file format and settings of the formatter), or not at all (when generating an RFC). This element appears as child element of: <annotation> (), <c> (), <postamble> (), <preamble> (), and <t> (). Content model: only text content.

Document-wide unique identifier for this comment. The processor will auto-generate an identifier when none is given. The value needs to be a valid XML "Name" (), additionally constrained to US-ASCII characters ().

Holds the "source" of a comment, such as the name or the initials of the person who made the comment.

Provides information about the publication date. Note that this element is used both for the boilerplate of the document being produced, and also inside bibliographic references. In the first case, it defines the publication date, which, when producing Internet-Drafts, will be used for computing the expiration date (see ). When "year", "month" or "day" are left out, the processor will attempt to use the current system date if the attributes that are specified do match the system date. Note that in the first case, month names need to match the full (English) month name ("January", "February", "March", "April", "May, "June", "July", "August", "September", "October", "November", or "December") in order for expiration calculations to work (some implementations might support additional formats, though). In the second case, the date information can have prose text for the month or year. For example, vague dates (year="ca. 2000"), date ranges (year="2012-2013") non-specific months (month="Second quarter") and so on, are allowed. This element appears as child element of: <front> (). Content model: this element does not have any contents.

Day of publication; this is a number.

Month of publication; this is the English name of the month.

Year of publication.

Provides an email address. The value is expected to be the scheme-specific part of a "mailto" URI (so does not include the prefix "mailto:"). See for details. This element appears as child element of: <address> (). Content model: only text content.

Represents an "external" link (as specified in the "target" attribute). If the element has text content, that content will be used. Otherwise, the value of the target attribute will be inserted in angle brackets (). This element appears as child element of: <annotation> (), <c> (), <postamble> (), <preamble> (), and <t> (). Content model: only text content.

URI of the link target (see ).

Represents the phone number of a fax machine. The value is expected to be the scheme-specific part of a "tel" URI (so does not include the prefix "tel:"), using the "global numbers" syntax. See for details. This element appears as child element of: <address> (). Content model: only text content.

This element is used to represent a figure, consisting of an optional preamble, the actual figure, an optional postamble, and an optional title. This element appears as child element of: <section> (), and <t> (). Content model: In this order: Optional <iref> elements () One optional <preamble> element () One <artwork> element () One optional <postamble> element ()

Used to change the alignment of <preamble> and <postamble>. Note: does not affect title or <artwork> alignment. Allowed values: "left" (default) "center" "right"

Duplicates functionality available on <artwork>; avoid it.

Document-wide unique identifier for this figure. Furthermore, the presence of this attribute causes the figure to be numbered. The value needs to be a valid XML "Name" ().

Duplicates functionality available on <artwork>; avoid it.

Duplicates functionality available on <artwork>; avoid it.

Figures that have an "anchor" attribute will automatically get an autogenerated title (such as "Figure 1"), even if the "title" attribute is absent. Setting this attribute to "true" will prevent this. Allowed values: "true" "false" (default)

The title for the figure; this usually appears on a line after the figure.

Duplicates functionality available on <artwork>; avoid it.

Provides a link to an additional format variant for a reference. Note that these additional links are neither used in published RFCs, nor supported by all tools. If the goal is to provide a single URI for a reference, the "target" attribute on <reference> can be used instead. This element appears as child element of: <reference> (). Content model: this element does not have any contents.

Octet length of linked-to document.

URI of document.

The type of the linked-to document, such as "TXT", "HTML", or "PDF".

Represent the "front matter": metadata (such as author information), abstract, and additional notes. This element appears as child element of: <reference> (), and <rfc> (). Content model: In this order: One <title> element () One or more <author> elements () One <date> element () Optional <area> elements () Optional <workgroup> elements () Optional <keyword> elements () One optional <abstract> element () Optional <note> elements ()

Provides terms for the document's index. Index entries can be either single items (when just the "item" attribute is given) or nested items (by specifying "subitem" as well).

For instance: <iref item="Grammar" subitem="item"/> will produce an index entry for "Grammar, item".

This element appears as child element of: <annotation> (), <c> (), <figure> (), <postamble> (), <preamble> (), <section> (), and <t> (). Content model: this element does not have any contents.

The item to include.

Setting this to "true" declares the occurrence as "primary", which might cause it to be highlighted in the index. Allowed values: "true" "false" (default)

The subitem to include.

Specifies a keyword applicable to the document. Note that each element should only contain a single keyword; for multiple keywords, the element can simply be repeated. Keywords are used both in the RFC Index and in the metadata of generated document formats. This element appears as child element of: <front> (). Content model: only text content.

Delineates a text list. Each list item is represented by a <t> element. The vocabulary currently does not directly support list items consisting of multiple paragraphs; if this is needed, <vspace> () can be used as workaround. This element appears as child element of: <t> (). Content model: One or more <t> elements ()

This attribute holds a token that serves as an identifier for a counter. The intended use is continuation of lists. Note that this attribute functions only when the style attribute is using the "format..." syntax (); otherwise, it is ignored.

For list styles with potentially wide labels, this attribute can override the default indentation level, measured in characters. Note that it only affects style with variable-width labels ("format..." and "hanging", see below), and it may not affect formats in which the list item text appears below the label.

This attribute is used to control the display of a list. The value of this attribute is inherited by any nested lists that do not have this attribute set. It may be set to: "empty" (or not set) For unlabeled list items; it can also be used for indentation purposes (this is the default value). "hanging" For lists where the items are labeled with a piece of text. The label text is specified in the 'hangText' attribute of the <t> element (). "letters" For ordered lists using letters as labels (lowercase letters followed by a period; after "z", it rolls over to a two-letter format). For nested lists, processors usually flip between uppercase and lowercase. "numbers" For ordered lists using numbers as labels. "symbols" For unordered (bulleted) lists. The style of the bullets is chosen automatically be the processor (some implementations allow overriding the default using a processing instruction). And, finally: "format ..." For lists with customized labels, consisting of fixed text and an item counter in various formats. The value is a free-form text that allows counter values to be inserted using a "percent-letter" format. For instance, "[REQ%d]" generates labels of the form "[REQ1]", where "%d" inserts the item number as decimal number. The following formats are supported: lowercase letters (a, b, c, etc.) uppercase letters (A, B, C, etc.) decimal numbers (1, 2, 3, etc.) lowercase Roman numerals (i, ii, iii, etc.) uppercase Roman numerals (I, II, III, etc.) represents a percent sign Other formats are reserved for future use.

Represents the main content of the document. This element appears as child element of: <rfc> (). Content model: One or more <section> elements ()

Creates an unnumbered section that appears after the abstract. It is usually used for additional information to reviewers (working group information, mailing list, ...), or for additional publication information such as "IESG Notes". This element appears as child element of: <front> (). Content model: One or more <t> elements ()

The title of the note.

Specifies the affiliation of an author. This information appears in both the "Author's Address" section and on the front page (see for more information). If the value is long, an abbreviated variant can be specified in the "abbrev" attribute. This element appears as child element of: <author> (). Content model: only text content.

Abbreviated variant.

Represents a phone number. The value is expected to be the scheme-specific part of a "tel" URI (so does not include the prefix "tel:"), using the "global numbers" syntax. See for details. This element appears as child element of: <address> (). Content model: only text content.

Contains child elements providing postal information. This element appears as child element of: <address> (). Content model: In this order: One or more <street> elements () In any order: <city> elements () <region> elements () <code> elements () <country> elements ()

Gives text that appears at the bottom of a figure or table. This element appears as child element of: <figure> (), and <texttable> (). Content model: In any order: Text <xref> elements () <eref> elements () <iref> elements () <cref> elements () <spanx> elements ()

Gives text that appears at the top of a figure or table. This element appears as child element of: <figure> (), and <texttable> (). Content model: In any order: Text <xref> elements () <eref> elements () <iref> elements () <cref> elements () <spanx> elements ()

Represents a bibliographical reference. This element appears as child element of: <references> (). Content model: In this order: One <front> element () Optional <seriesInfo> elements () Optional <format> elements () Optional <annotation> elements ()

Document-wide unique identifier for this reference. Usually, this will be used both to "label" the reference in the references section, and as an identifier in links to this reference entry. The value needs to be a valid XML "Name" (), additionally constrained to US-ASCII characters ().

Holds the URI for the reference. Note that depending on the <seriesInfo> element, a URI might not be needed, nor desirable, as it can be automatically generated (for instance, for RFCs).

Contains a set of bibliographical references. In the early days of the RFC series, there was only one "References" section per RFC. This convention was later changed to group references into two sets, "Normative" and "Informative" as described in ). This vocabulary supports the split with the "title" attribute. This element appears as child element of: <back> (). Content model: One or more <reference> elements ()

Provides the title for the References section (defaulting to "References"). In general, the title should be either "Normative References" or "Informative References".

Provides the region name in a postal address. This element appears as child element of: <postal> (). Content model: only text content.

This is the root element of the xml2rfc vocabulary. Processors distinguish between RFC mode ("number" attribute being present) and Internet-Draft mode ("docName" attribute being present): it is invalid to specify both. Setting neither "number" nor "docName" can be useful for producing other types of document but is out-of-scope for this specification. Content model: In this order: One <front> element () One <middle> element () One optional <back> element ()

Document category (see ). Allowed values: "std" "bcp" "info" "exp" "historic"

Affects the generated boilerplate. See for more information. Allowed values: "no" "yes"

For Internet-Drafts, this specifies the draft name (which appears below the title). A processor should give an error if both the "docName" and "number" attributes are given in the <rfc> element. Note that the file extension is not part of the draft, so in general it should end with the current draft number ("-", plus two digits). Furthermore, it is good practice to disambiguate current editor copies from submitted drafts (for instance, by replacing the draft number with the string "latest"). See for further information.

Represents the Intellectual Property status of the document. See for details. Allowed values: "full2026" "noDerivativeWorks2026" "none" "full3667" "noModification3667" "noDerivatives3667" "full3978" "noModification3978" "noDerivatives3978" "trust200811" "noModificationTrust200811" "noDerivativesTrust200811" "trust200902" "noModificationTrust200902" "noDerivativesTrust200902" "pre5378Trust200902"

Identifies a single section within the document (by its 'anchor' attribute) for which extraction "as-is" is explicitly allowed (this is only relevant for historic values of the "ipr" attribute).

The number of the RFC to be produced. A processor should give an error if both the "docName" and "number" attributes are given in the <rfc> element.

A comma-separated list of RFC numbers or Internet-Draft names. Processors ought to parse the attribute value, so that incorrect references can be detected and, depending on output format, hyperlinks can be generated. Also, the value ought to be reformatted to insert whitespace after each comma if not already present.

When producing a document within document series (such as "STD"): the number within that series.

The document stream. See for details. Allowed values: "IETF" (default) "IAB" "IRTF" "independent"

A comma-separated list of RFC numbers or Internet-Draft names. Processors ought to parse the attribute value, so that incorrect references can be detected and, depending on output format, hyperlinks can be generated. Also, the value ought to be reformatted to insert whitespace after each comma if not already present.

The natural language used in the document (defaults to "en"). See for more information.

Represents a section (when inside a <middle> element) or an appendix (when inside a <back> element). Sub-sections are created by nesting <section> elements inside <section> elements. This element appears as child element of: <back> (), <middle> (), and <section> (). Content model: In this order: In any order: <t> elements () <figure> elements () <texttable> elements () <iref> elements () Optional <section> elements ()

Document-wide unique identifier for this section. The value needs to be a valid XML "Name" ().

The title of the section.

Determines whether the section is included in the Table Of Contents. The processor usually has defaults for whether a Table Of Contents will be produced at all, and sections of which maximal depth will be included (frequently: 3). "include" and "exclude" allow overriding the processor's default behavior for the element they are specified on (they do not affect nested elements). Allowed values: "include" "exclude" "default" (default)

Specifies the document series in which this document appears, and also specifies an identifier within that series. This element appears as child element of: <reference> (). Content model: this element does not have any contents.

The name of the series. The following names trigger specific processing (such as for auto-generating links, and adding descriptions such as "work in progress"): "BCP", "FYI", "Internet-Draft", "RFC", and "STD".

The identifier within the series specified by the "name" attribute. For BCPs, FYIs, RFCs, and STDs this is the number within the series. For Internet-Drafts, it is the full draft name (ending with the two-digit version number).

Wraps a piece of text, indicating special formatting styles. When generating plain text, processors usually emulate font changes using characters such as "*" and "_". The following styles are defined: Simple emphasis (this is the default). Strong emphasis. "Verbatim" text (usually displayed usign a monospaced font face). This element appears as child element of: <annotation> (), <c> (), <postamble> (), <preamble> (), and <t> (). Content model: only text content.

The style to be used (defaults to "emph").

Determines whitespace handling. According to the DTD, the default value is "preserve". Tests however show that it doesn't have any effect on processing; thus this attribute will be removed in future versions of the vocabulary. See also . Allowed values: "default" "preserve" (default)

Provides a street address. This element appears as child element of: <postal> (). Content model: only text content.

Contains a paragraph of text. This element appears as child element of: <abstract> (), <list> (), <note> (), and <section> (). Content model: In any order: Text <list> elements () <figure> elements () <xref> elements () <eref> elements () <iref> elements () <cref> elements () <spanx> elements () <vspace> elements ()

Document-wide unique identifier for this paragraph. The value needs to be a valid XML "Name" ().

Holds the label ("hanging text") for items in lists using the "hanging" style (see ).

Contains a table, consisting of an optional preamble, a header line, rows, an optional postamble, and an optional title. The number of columns in the table is determined by the number of <ttcol> elements. The number of rows in the table is determined by the number of <c> elements divided by the number of columns. There is no requirement that the number of <c> elements be evenly divisible by the number of columns. This element appears as child element of: <section> (). Content model: In this order: One optional <preamble> element () One or more <ttcol> elements () Optional <c> elements () One optional <postamble> element ()

Determines the horizontal alignment of the table. Allowed values: "left" "center" (default) "right"

Document-wide unique identifier for this table. Furthermore, the presence of this attribute causes the table to be numbered. The value needs to be a valid XML "Name" ().

Selects which borders should be drawn, where "all" means borders around all table cells, "full" is like "all" except no horizontal lines between table rows (except below the column titles), "headers" adds just a separator between column titles and rows, and "none" means no borders at all. Allowed values: "all" "none" "headers" "full" (default)

Tables that have an "anchor" attribute will automatically get an autogenerated title (such as "Table 1"), even if the "title" attribute is absent. Setting this attribute to "true" will prevent this. Allowed values: "true" "false" (default)

The title for the table; this usually appears on a line below the table body.

Represents the document title. When this element appears in the <front> element of the current document, the title might also appear in page headers or footers. If it's long (~40 characters), the "abbrev" attribute is used to specify an abbreviated variant. This element appears as child element of: <front> (). Content model: only text content.

Specifies an abbreviated variant of the document title.

Contains a column heading in a table. This element appears as child element of: <texttable> (). Content model: only text content.

Determines the horizontal alignment within the table column. Allowed values: "left" (default) "center" "right"

The desired column width (as integer 0..100 followed by "%").

Contains a web address associated with the author. The contents should be a valid URI (see ). This element appears as child element of: <address> (). Content model: only text content.

This element can be used to force the inclusion of a single line break or multiple blank lines. Note that this is a purely presentational element and thus its use ought to be avoided. This element appears as child element of: <t> (). Content model: this element does not have any contents.

Number of blank lines to be inserted, where "0" indicates a single line break (defaults to "0"). For paged output formats, no additional blank lines should be generated after a page break.

This element is used to specify the Working Group (IETF) or Research Group (IRTF) from which the document originates, if any. The recommended format is the official name of the Working Group (with some capitalization). In Internet-Drafts, this is used in the upper left corner of the boilerplate, replacing the "Network Working Group" string. Formatting software can append the words "Working Group" or "Research Group", depending on the "submissionType" property on the <rfc> element (). This element appears as child element of: <front> (). Content model: only text content.

Inserts a reference to a different part of a document. The generated text depends on whether the <xref> is empty (in which case the processor will try to generate a meaningful text fragment), and the nature of the referenced document part. Any element that allows the "anchor" attribute can be referenced, however there are restrictions with respect to the text content being generated. For instance, a <t> can be a reference target, however, because paragraphs are not (visibly) numbered, the author will have to make sure that the prose is sufficient for a reader to understand what is being referred to. This needs to be expanded with examples and with a discussion how the autogenerated text differs when <xref> is not empty This element appears as child element of: <annotation> (), <c> (), <postamble> (), <preamble> (), and <t> (). Content model: only text content.

This attribute is used to control the format of the generated reference text. "counter" Inserts a counter, such as the number of a section, figure, or table. "default" Inserts a text fragment that describes the referenced part completely, such as "Section 2", "Table 4", or "[XML]". "none" There will be no auto-generated text. "title" Inserts a title for the referenced element (usually obtained from the referenced element's "title" attribute; some processors also use the <title> child element or a <reference> target). Allowed values: "counter" "title" "none" "default" (default)

Unused. It's unclear what the purpose of this attribute is; processors seem to ignore it and it never was documented. Allowed values: "true" "false" (default)

Identifies the document component being referenced. The value needs to match the value of the "anchor" attribute of another element in the document.

Text in XML can not use the literal characters "<" and "&", as they have special meaning to the XML processor (starting entities, elements, etc.). Usually, these characters will need to be substituted by "&lt;" and "&amp;" (see ). ">" does not require escaping, unless it appears in the sequence "]]>" (which indicates the end of a CDATA section, see below). Escaping the indivual characters can be a lot of work (when done manually), and also messes up alignment in artwork. Another approach to escaping is to use CDATA sections (). Within these, no further escaping is needed, except when the "end-of-CDATA" marker needs to be used (in that case, the CDATA section needs to be closed, and a new one needs to be started).

Although the current RFC format does not allow non-ASCII Unicode characters (), some of them can be used to enforce certain behaviors of formatters. For instance: non-breaking space (U+00A0) Represents a space character where no line break should happen. This is frequenly used in titles (by excluding certain space characters from the line breaking algorithm, the processor will use the remaining whitespace ocurrences for line breaks). non-breaking hyphen (U+2011) Similarly, this represents a hyphen character where nevertheless no line breaking ought to occur. word joiner (U+2060) Also called "zero width non-breaking space" — can be used to disallow line breaking between two non-whitespace characters. Note that in order to use these characters by name, they need to be declared either in the Document Type Definition (DTD, ), or in the "internal subset" (), like that:

<?xml version="1.0"?> <!DOCTYPE rfc [ <!-- declare nbsp and friends --> <!ENTITY nbsp "&#xa0;"> <!ENTITY nbhy "&#x2011;"> <!ENTITY wj "&#x2060;"> ]>

This version of the vocabulary does not support an inclusion mechanism on it's own — thus, a document always needs to be self-contained. That being said, some processors do support file inclusion using processing instructions (). Furthermore, XML itself allows inclusion of external content using the "internal subset" (). Unfortunately, this requires declaring the external data in the DTD upfront.

For instance: <?xml version="1.0"?> <!DOCTYPE rfc [ <!-- allow later RFC2616 reference using "&rfc2616;" --> <!-- the data will be fetched from xml.resource.org --> <!ENTITY rfc2616 PUBLIC "http://xml.resource.org/public/rfc/bibxml/reference.RFC.2616.xml"> ]> ...declares the entity "rfc2616", which then can be used in the "references" section:

<references> &rfc2616; </references>

Note that this mechanism only works for well-formed XML fragments; thus any plain text that would need to be escaped in XML can't be included as-is.

This format is based on , thus does not have any issues representing arbitrary Unicode characters in text content. However, the current canonical RFC format is restricted to US-ASCII characters (). Future versions are likely to relax this role, and it is expected that the vocabulary will be extended so that US-ACSII alternatives can be provided when that makes sense (for instance, in contact information).

The "name" attribute on the <artwork> element () can be used to derive a filename for saving to a local file system. Trusting this kind of information without pre-processing is a known security risk; see for more information. Furthermore, the nature of XML, plus vocabulary features such as typed artwork, make it attractive to extract content from documents for further processing, such for the purpose of checking syntax, or computing/verifying examples. In the latter case, care needs to be taken that only trusted content is processed. Furthermore, all security considerations related to XML processing are relevant as well (see ).

IANA maintains the registry of Internet media types at . This document serves as the specification for the Internet media type "application/rfc+xml". The following is to be registered with IANA. application rfc+xml There are no required parameters. "charset": This parameter has identical semantics as the charset parameter of the "application/xml" media type specified in . Identical to those of "application/xml" as described in . As defined in . In addition, as this media type uses the "+xml" convention, it inherits the security considerations described in . N/A This specification. Applications that either transform xml2rfc to output formats such as plain text or HTML, plus additional analysis tools. The "anchor" attribute is used for assigning document-wide unique identifiers that can be uses as shorthand pointers, as described in . None. As specified for "application/xml" in . .xml or .rfcxml when disambiguation from other XML files is needed TEXT See Authors Section. COMMON N/A See Authors Section. RFC Series Editor (rse@rfc-editor.org)

Thanks to everybody who reviewed this document and provided feedback and/or specification text, in particular Brian Carpenter, Tony Hansen, Paul Hoffman, Henrik Levkowetz, Alice Russo, Tom Taylor, Jim Schaad, and Nico Williams. We also thank Marshall T. Rose for both the original design and the reference implementation of the "xml2rfc" formatter.

Latest version available at . PDF version:

jjc@jclark.com

RFC Editor IETF Trust IETF Trust IETF Trust The Unicode Consortium American National Standards Institute Latest version available at . Latest version available at . Thai Open Source Software Center Ltd Downloads: .

For RFCs, the category determines the "maturity level" (see ). The allowed values are "std" for "Standards Track", "bcp" for "BCP", "info" for "Informational", "exp" for "Experimental", and "historic" for "Historic". For Internet-Drafts, the category attribute is not needed, but will appear on the front page as "Intended Status". Supplying this information can be useful to reviewers.

This attribute value can take a long list of values, each of which describes an IPR policy for the document. This attribute's values are not the result of a grand plan, but remain simply for historic reasons. Of these values, only a few are currently in use; all others are supported by the various tools for backwards compatibility with old source files. Note: some variations of the boilerplate are selected based on the document's date; therefore it is important to specify the "year", "month" and "day" attributes of the <date> element when archiving the XML source of an Internet-Draft on the day of submission. Disclaimer: THIS ONLY PROVIDES IMPLEMENTATION INFORMATION. IF YOU NEED LEGAL ADVICE, PLEASE CONTACT A LAWYER. For further information, refer to . For the current "Status Of This Memo" text, the submissionType attribute determines whether a statement about "Code Components" is inserted (which is the case for the value "IETF", which is the default). Other values, such as "independent", suppress this part of the text.

The name for these values refers to the "IETF TRUST Legal Provisions Relating to IETF Documents", sometimes simply called the "TLP, that went into effect on February 15, 2009 (). Updates to this document were published on September 12, 2009 () and on December 28, 2009 (), modifying the license for code components (see for further information). The actual text is located in Section 6 ("Text To Be Included in IETF Documents") of these documents. The tools will automatically produce the "correct" text depending on the document's date information (see above): TLPstarting with publication date 2009-11-01 2010-04-01

This should be the default, unless one of the more specific '*trust200902' values is a better fit. It produces the text in Sections 6.a and 6.b of the TLP.

This produces additional text from Section 6.c.i of the TLP: This document may not be modified, and derivative works of it may not be created, except to format it for publication as an RFC or to translate it into languages other than English. Note: this clause is incompatible with RFCs that are published on the Standards Track.

This produces the additional text from Section 6.c.ii of the TLP: This document may not be modified, and derivative works of it may not be created, and it may not be published except as an Internet-Draft. Note: this clause is incompatible with RFCs.

This produces the additional text from Section 6.c.iii of the TLP, frequently called the "pre-5378 escape clause": This document may contain material from IETF Documents or IETF Contributions published or made publicly available before November 10, 2008. The person(s) controlling the copyright in some of this material may not have granted the IETF Trust the right to allow modifications of such material outside the IETF Standards Process. Without obtaining an adequate license from the person(s) controlling the copyright in such materials, this document may not be modified outside the IETF Standards Process, and derivative works of it may not be created outside the IETF Standards Process, except to format it for publication as an RFC or to translate it into languages other than English. See Section 4 of for further information about when to use this value. Note: this text appears under "Copyright Notice", unless the document was published before November 2009, in which case it appears under "Status Of This Memo".

The attribute values "trust200811", "noModificationTrust200811" and "noDerivativesTrust200811" are similar to their "trust200902" counterparts, except that they use text specified in .

The attribute values "full3978", "noModification3978" and "noDerivatives3978" are similar to their counterparts above, except that they use text specified in RFC 3978 (March 2005).

The attribute values "full3667", "noModification3667" and "noDerivatives3667" are similar to their counterparts above, except that they use text specified in RFC 3667 (February 2004).

The attribute values "full2026" and "noDerivativeWorks2026" are similar to their counterparts above, except that they use text specified in RFC 2026 (October 1996). The special value "none" was also used back then, and denied the IETF any rights beyond publication as Internet-Draft.

See for a comparison of the schemata.

The <appendix> element has been removed; to generate an appendix, place a <section> inside <back>.

Many attributes have lost their "default" value; this is to avoid having document semantics differ based on whether a DTD was specified and evaluated. Processors will handle absent values the way the default value was specified before.

<artwork>: Has a set of new attributes: "name", "type", "src", "align", "alt", "width", and "height". () <author>: The <organization> element is now optional. The "role" attribute was added. () <country>: The requirement to use ISO 3166 codes was removed. () <date>: All attributes are now optional. () <figure>: Has a set of new attributes: "suppress-title", "src", "align", "alt", "width", and "height". () <iref>: Has a new "primary" attribute. () <list>: The "style" attribute isn't restricted to a set of enumerated values anymore. The "hangIndent" and "counter" attributes have been added. () <rfc>: The "ipr" attribute has gained additional values. The attributes "consensus", "iprExtract", "submissionType", and "xml:lang" have been added. () <reference>: <annotation> allows adding prose to a reference. The "anchor" attribute has been made mandatory. () <references>: Can now appear multiple times, and carry a "title" attribute (so that normative and informative references can be split). () <section>: The new "toc" attribute controls whether it will appear in the Table Of Contents. <iref> can now appear as direct child element. () <t>: The "anchor" attribute can now be used as well, however there are restrictions on how they can be referred to. ()

The following elements have been added: <annotation> (), <c> (), <cref> (), <format> (), <spanx> (), <texttable> ().

namespace a = "http://relaxng.org/ns/compatibility/annotations/1.0" rfc = element rfc { attribute number { text }?, [ a:defaultValue = "" ] attribute obsoletes { text }?, [ a:defaultValue = "" ] attribute updates { text }?, attribute category { "std" | "bcp" | "info" | "exp" | "historic" }?, attribute consensus { "no" | "yes" }?, attribute seriesNo { text }?, attribute ipr { "full2026" | "noDerivativeWorks2026" | "none" | "full3667" | "noModification3667" | "noDerivatives3667" | "full3978" | "noModification3978" | "noDerivatives3978" | "trust200811" | "noModificationTrust200811" | "noDerivativesTrust200811" | "trust200902" | "noModificationTrust200902" | "noDerivativesTrust200902" | "pre5378Trust200902" }?, attribute iprExtract { xsd:IDREF }?, [ a:defaultValue = "IETF" ] attribute submissionType { "IETF" | "IAB" | "IRTF" | "independent" }?, attribute docName { text }?, [ a:defaultValue = "en" ] attribute xml:lang { text }?, front, middle, back? } front = element front { title, author+, date, area*, workgroup*, keyword*, abstract?, note* } title = element title { attribute abbrev { text }?, text } author = element author { attribute initials { text }?, attribute surname { text }?, attribute fullname { text }?, attribute role { "editor" }?, organization?, address? } organization = element organization { attribute abbrev { text }?, text } address = element address { postal?, phone?, facsimile?, email?, uri? } postal = element postal { street+, (city | region | code | country)* } street = element street { text } city = element city { text } region = element region { text } code = element code { text } country = element country { text } phone = element phone { text } facsimile = element facsimile { text } email = element email { text } uri = element uri { text } date = element date { attribute day { text }?, attribute month { text }?, attribute year { text }?, empty } area = element area { text } workgroup = element workgroup { text } keyword = element keyword { text } abstract = element abstract { t+ } note = element note { attribute title { text }, t+ } middle = element middle { section+ } section = element section { attribute anchor { xsd:ID }?, attribute title { text }, [ a:defaultValue = "default" ] attribute toc { "include" | "exclude" | "default" }?, (t | figure | texttable | iref)*, section* } t = element t { attribute anchor { xsd:ID }?, attribute hangText { text }?, (text | \list | figure | xref | eref | iref | cref | spanx | vspace)* } \list = element list { attribute style { text }?, attribute hangIndent { text }?, attribute counter { text }?, t+ } xref = element xref { attribute target { xsd:IDREF }, [ a:defaultValue = "false" ] attribute pageno { "true" | "false" }?, [ a:defaultValue = "default" ] attribute format { "counter" | "title" | "none" | "default" }?, text } eref = element eref { attribute target { text }, text } iref = element iref { attribute item { text }, [ a:defaultValue = "" ] attribute subitem { text }?, [ a:defaultValue = "false" ] attribute primary { "true" | "false" }?, empty } cref = element cref { attribute anchor { xsd:ID }?, attribute source { text }?, text } spanx = element spanx { [ a:defaultValue = "preserve" ] attribute xml:space { "default" | "preserve" }?, [ a:defaultValue = "emph" ] attribute style { text }?, text } vspace = element vspace { [ a:defaultValue = "0" ] attribute blankLines { text }?, empty } figure = element figure { attribute anchor { xsd:ID }?, [ a:defaultValue = "" ] attribute title { text }?, [ a:defaultValue = "false" ] attribute suppress-title { "true" | "false" }?, attribute src { text }?, [ a:defaultValue = "left" ] attribute align { "left" | "center" | "right" }?, [ a:defaultValue = "" ] attribute alt { text }?, [ a:defaultValue = "" ] attribute width { text }?, [ a:defaultValue = "" ] attribute height { text }?, iref*, preamble?, artwork, postamble? } preamble = element preamble { (text | xref | eref | iref | cref | spanx)* } artwork = element artwork { [ a:defaultValue = "preserve" ] attribute xml:space { "default" | "preserve" }?, [ a:defaultValue = "" ] attribute name { text }?, [ a:defaultValue = "" ] attribute type { text }?, attribute src { text }?, [ a:defaultValue = "left" ] attribute align { "left" | "center" | "right" }?, [ a:defaultValue = "" ] attribute alt { text }?, [ a:defaultValue = "" ] attribute width { text }?, [ a:defaultValue = "" ] attribute height { text }?, text* } postamble = element postamble { (text | xref | eref | iref | cref | spanx)* } texttable = element texttable { attribute anchor { xsd:ID }?, [ a:defaultValue = "" ] attribute title { text }?, [ a:defaultValue = "false" ] attribute suppress-title { "true" | "false" }?, [ a:defaultValue = "center" ] attribute align { "left" | "center" | "right" }?, [ a:defaultValue = "full" ] attribute style { "all" | "none" | "headers" | "full" }?, preamble?, ttcol+, c*, postamble? } ttcol = element ttcol { attribute width { text }?, [ a:defaultValue = "left" ] attribute align { "left" | "center" | "right" }?, text } c = element c { (text | xref | eref | iref | cref | spanx)* } back = element back { references*, section* } references = element references { [ a:defaultValue = "References" ] attribute title { text }?, reference+ } reference = element reference { attribute anchor { xsd:ID }, attribute target { text }?, front, seriesInfo*, format*, annotation* } seriesInfo = element seriesInfo { attribute name { text }, attribute value { text }, empty } format = element format { attribute target { text }?, attribute type { text }, attribute octets { text }?, empty } annotation = element annotation { (text | xref | eref | iref | cref | spanx)* } start = rfc

(This schema was derived from version 1.3.6 of the xml2rfc DTD ('Document Type Definition', ), available from ).

The validity of XML files can be checked with any tool that supports Relax NG (). The reference implementation is the Java-based, open sourced "JING" (). To use JING, download the latest ZIP file from the downloads page (currently ), extract the archive, copy "jing.jar" from the "bin" folder, and make sure Java is installed). To check a file "test.xml" using the RNC file "schema.rnc", run (from a command line prompt):

java -jar jing.jar -c schema.rnc test.xml

In good Unix tradition, no output means the file is valid.

Discussion of "v3" changes takes place on the rfc-interest mailing list (rfc-interest@rfc-editor.org), which has its home page at . See also for a related Wiki page.

If contact information is changed to allow non-ASCII characters: add a place for a ASCII fallback (probably just for the author names). The content model for <postal> ought to be more strict to allow at most one of <city>, <region>, <code>, and <country>. It should be possible to have multiple <email> and <uri> elements (see also ). <facsimile> looks outdated, while a container for IM (messaging) URIs is missing. Maybe this area needs to be aligned with vCard. hints at a "Contributors" Section that could supply contact information similar to the one in the auto-generated "Authors' Address" Section. Consider how to capture contributor contact information (probably not using <author> to avoid confusion). Furthermore, consider ways to augment the contact information section with prose.

Cleanup the set of overlapping attributes between <figure> and <artwork>. For artwork that consists of a sequence of items (such as messages in a protocol example), it would be good if a <figure> element could contain multiple <artwork> elements (to assist code to find good places for page breaks). Extend <figure> to support different types of artwork (such as by specifying certain type attribute values, see ), and also avoid having to markup code (such as ABNF) as "artwork". It would be good if "code components" could be marked as such. Finally, even in preformatted text use of markup could be useful to support (a) references, or (b) highlighting the important bits ().

Extend <xref> so that subsection/anchors can be specified (see ). Remove the "pageno" attribute which seems to be both undocumented and non-functional.

Allow multiple paragraphs in list items; eliminating the need to use <vspace> — this could be achieved by adding a list item container element ("<lt>", see and ). Add support for a "dictionary" style; eliminating the need to combine "hanging" with <vspace> to force new lines (see thread around ).

Allow overriding the "anchor" attribute of an included <reference> element. Add a way to add prose to a reference that avoids abuse of <seriesInfo>. Allow <reference>s that identify a document set such as a BCP. Deprecate or remove the <format> element; right now it's not used for the generation of the plain text document anyway. The "anchor" attribute is optional because it is not needed when using numeric references (symrefs processing instruction), and the reference actually is not in use. This is an edge case that doesn't need special support in the vocabulary and thus should be removed.

When this vocabulary becomes the canonical RFC format, it will need to be able to capture all generated information, such as section/figure/table numbers, plus any auto-generated boilerplate (copyright statements etc.).

Extend the concept of language tagging to at least examples and contact information to address potential japanese/chinese font confusion. Provide a way to indicate the intended level on the standards track. Include feedback information in a way so that generated documents can provide usable feedback links (see ).

As discussed in , file inclusion currently uses out-of-the-box XML mechanisms or processor-specific directives. We need to decide whether the vocabulary should have its own inclusion mechanism, or whether it would be better to use a generic solution such as instead.

Make the <date> element optional; all of its content is optional already. <spanx> has both a weird whitespace model ("preserve") and problematic styling. Consider to deprecate it in favor of elements such as <b>, <i>, and <tt>. Indented paragraphs currently can be created by abusing the <list>. It would be good to have a special element for this purpose. Provide a special element for inserting block quotes (). The content model for <cref> should be extended to allow more flow elements, such as <xref> and <eref>. Section titles should really be elements, not attributes (this would allow them to contain markup). Text tables are currently very constrained. For instance, it would be good if alignment of headers and table cells could be de-coupled (). Counters are currently restricted to lists, figures, and tables. Maybe there should be a generic mechanism that is not directly tied to other elements ().