Internet Draft                                          Paul Hoffman
draft-hoffman-i18n-terms-08.txt                           IMC & VPNC
October 4, 2002
Expires in six months

          Terminology Used in Internationalization in the IETF

Status of this memo

This document is an Internet-Draft and is in full conformance with all
provisions of Section 10 of RFC 2026.

Internet-Drafts are working documents of the Internet Engineering Task
Force (IETF), its areas, and its working groups. Note that other groups
may also distribute working documents as Internet-Drafts.

Internet-Drafts are draft documents valid for a maximum of six months
and may be updated, replaced, or obsoleted by other documents at any
time. It is inappropriate to use Internet-Drafts as reference material
or to cite them other than as "work in progress."

The list of current Internet-Drafts can be accessed at
http://www.ietf.org/ietf/1id-abstracts.txt

The list of Internet-Draft Shadow Directories can be accessed at
http://www.ietf.org/shadow.html.

Abstract

This document provides a glossary of terms used in the IETF when
discussing internationalization. The purpose is to help frame
discussions of internationalization in the various areas of the IETF and
to help introduce the main concepts to IETF participants.


. Introduction

As [RFC2277] summarizes: "Internationalization is for humans. This means
that protocols are not subject to internationalization; text strings
are." Many protocols throughout the IETF use text strings that are
entered by, or are visible to, humans. It should be possible for anyone
to enter or read these text strings, which means that Internet users
must be able to be enter text in typical input methods and displayed in
any human language. Further, text containing any character should be
able to be passed between Internet applications easily. This is the
challenge of internationalization.

1.1 About this document

Internationalization is discussed in many working groups of the IETF.
However, few working groups have internationalization experts. When
designing or updating protocols, the question often comes up "should we
internationalize this" (or, more likely, "do we have to internationalize
this").

This document gives an overview of internationalization as it applies to
IETF standards work by lightly covering the many aspects of
internationalization and the vocabulary associated with those topics. It
is not meant to be a complete description of internationalization. The
definitions in this document are not normative for IETF standards;
however, they are useful and standards may make non-normative reference
to this document after it becomes an RFC. Some of the definitions in
this document come from many earlier IETF documents and books.

As in many fields, there is disagreement in the internationalization
community on definitions for many words. The topic of language brings up
particularly passionate opinions for experts and non-experts alike. This
document attempts to define terms in a way that will be most useful to
the IETF audience.

This document uses definitions from many documents that have been
developed outside the IETF. The primary documents used are:
- ISO/IEC 10646 [ISOIEC10646]
- The Unicode Standard [UNICODE]
- W3C Character Model [CHARMOD]
- IETF RFCs, including [RFC2277]

In the body of this document, the source for the definition is shown in
angle brackets, such as "<ISOIEC10646>". Many definitions are
shown as "<NONE>", which means that the definitions were crafted
originally for this document. The angle bracket notation for the source
of definitions is different than the square bracket notation used
for references to documents, such as in the paragraph above; these
references are given in Section 9.

For some terms, there are commentary and examples after the
definitions. In those cases, the part before the angle brackets is the
definition that comes from the original source, and the part after the
angle brackets is commentary that is not a definition (such as
examples or further exposition).

Examples in this document use the notation for code points and names
from the Unicode Standard [UNICODE] and ISO/IEC 10646 [ISOIEC10646]. For
example, the letter "a" may be represented as either "U+0061" or "LATIN
SMALL LETTER A".


. Fundamental Terms

This section covers basic topics that are needed for almost anyone who
is involved with internationalization of IETF protocols.

language

A language is a way that humans interact. The use of language occurs in
many forms, the most common of which are speech, writing, and signing.
<NONE>

Some languages have a close relationship between the written and spoken
forms, while others have a looser relationship. [RFC3066] discusses
languages in more detail and provides identifiers for languages for use
in Internet protocols. Note that computer languages are explicitly
excluded from this definition.

script

A set of graphic characters used for the written form of one or more
languages. <ISOIEC10646>

Examples of scripts are Latin, Cyrillic, Greek, Arabic, and Han (the
ideographs used in writing Chinese, Japanese, and Korean). [RFC2277]
discusses scripts in detail.

It is common for internationalization novices to mix up the terms
"language" and "script".  This can be a problem in protocols that
differentiate the two. Almost all internationalized protocols deal with
scripts (the written systems) or characters, while fewer actually
deal with languages.

A single name can mean either a language or a script; for example,
"Arabic" is both the name of a language and the name of a script. In
fact, many scripts borrow their names from the names of languages.
Further, many scripts are used for many languages; for example, the
Russian and Bulgarian languages are written in the Cyrillic script. Some
languages can be expressed using different scripts; the Mongolian
language can be written in either the Mongolian and Cyrillic scripts,
and the Serbo-Croatian language is written using both the Latin and
Cyrillic scripts. Further, some languages are normally expressed with
more than one script at the same time; for example, the Japanese
language is normally expressed in the Kanji (Han), Katakana, and
Hiragana scripts in a single string of text.

character

A member of a set of elements used for the organization, control, or
representation of data. <ISOIEC10646>

There are at least three common definitions of the word "character":

- a general description of a text entity

- a unit of a writing system, often synonymous with "letter" or similar
terms

- the encoded entity itself

When people talk about characters, they are mostly using one of the
first two definitions. Standards, however, mostly use the third
definition, which is more strict.

A particular character is identified by its name, not by its shape. A
name may suggest a meaning, but the character may be used for
representing other meanings as well. A name may suggest a shape, but
that does not imply that only that shape is commonly used in print,
nor that the particular shape is associated only with that name.

coded character

A character together with its coded representation. <ISOIEC10646>

coded character set

A coded character set (CCS) is a set of unambiguous rules that
establishes a character set and the relationship between the characters
of the set and their coded representation. <ISOIEC10646>

character encoding form

A character encoding form is a mapping from a character set definition
to the actual code units used to represent the data. <UNICODE>

repertoire

The collection of characters included in a character set. Also
called a character repertoire. <UNICODE>

glyph

A glyph is an abstract form that represents one or more glyph images.
The term "glyph" is often a synonym for glyph image, which is the
actual, concrete image of a glyph representation having been rasterized
or otherwise imaged onto some display surface. In displaying character
data, one or more glyphs may be selected to depict a particular
character. These glyphs are selected by a rendering engine during
composition and layout processing. <UNICODE>

glyph code

A glyph code is a numeric code that refers to a glyph. Usually, the
glyphs contained in a font are referenced by their glyph code. Glyph
codes are local to a particular font; that is, a different font
containing the same glyphs may use different codes. <UNICODE>

transcoding

Transcoding is the process of converting text data from one character
encoding form to another. Transcoders work only at the level of
character encoding and do not parse the text. Note: Transcoding may
involve one-to-one, many-to-one, one-to-many or many-to-many mappings.
Because some legacy mappings are glyphic, they may not only be
many-to-many, but also discontinuous: thus XYZ may map to yxz. <CHARMOD>

In this definition, "many-to-one" means a sequence of characters mapped
to a single character. The "many" does not mean alternative characters
that map to the single character.

character encoding scheme

A character encoding scheme (CES) is a character encoding form plus byte
serialization. There are many character encoding schemes in Unicode,
such as UTF-8 and UTF-16. <UNICODE>

Some CESs are associated with a single CCS; for example, UTF-8 [RFC2279]
applies only to ISO/IEC 10646. Other CESs, such as ISO 2022, are
associated with many CCSs.

charset

A charset is a method of mapping a sequence of octets to a sequence of
abstract characters. A charset is, in effect, a combination of one or
more CCSs with a CES. Charset names are registered by the IANA according
to procedures documented in [RFC2278]. <NONE>

Many protocol definitions use the term "character set" in their
descriptions. The terms "charset" or "character encoding scheme" are
strongly preferred over the term "character set" because "character set"
has other definitions in other contexts and this can be confusing.

internationalization

In the IETF, "internationalization" means to add or improve the
handling of international information in a protocol. <NONE>

Many protocols that handle text only handle one script (often, the one
that contains the letters used in English text), or leave the question
of what character set is used up to local guesswork (which leads, of
course, to interoperability problems). Internationalizing such a
protocol allows the protocol to handle more scripts, hopefully all of
the ones useful in the world.

localization

The process of adapting an internationalized application platform or
application to a specific cultural environment. In localization, the
same semantics are preserved while the syntax may be changed.
[FRAMEWORK]

Localization is the act of tailoring an application for a different
language or script or culture. Some internationalized applications can
handle a wide variety of languages. Typical users only understand a
small number of languages, so the program must be tailored to interact
with users in just the languages they know.

The major work of localization is translating the user interface and
documentation, not dealing with localization of protocols. Localization
involves not only changing the language interaction, but also other
relevant changes such as display of numbers, dates, currency, and so on.
The better internationalized an application is, the easier it is to
localize it for a particular language and character encoding scheme.

Localization is rarely an IETF matter, and protocols that are merely
localized, even if they are serially localized for several locations,
are generally considered unsatisfactory for the global Internet.

Do not confuse "localization" with "locale", which is described in
Section 7 of this document.

i18n, l10n

These are abbreviations for "internationalization" and "localization".
<NONE>

"18" is the number of characters between the "i" and the "n" in
"internationalization", and "10" is the number of characters between the
"l" and the "n" in "localization".

multilingual

The term "multilingual" has many widely-varying definitions and thus is
not recommended for use in standards. Some of the definitions relate to
the ability to handle international characters; other definitions relate
to the ability to handle multiple charsets; and still others relate to
the ability to handle multiple languages. <NONE>

displaying and rendering text

To display text, a system puts characters on a visual display device
such as a screen or a printer. To render text, a system analyzes the
character input to determine how to display the text. The terms
"display" and "render" are sometimes used interchangeably. Note,
however,  that text might be rendered as audio and/or tactile output,
such as in systems that have been designed for people with visual
disabilities. <NONE>

Combining characters modify the display of the character (or, in some
cases, characters) that precede them. When rendering such text, the
display engine must either find the glyph in the font that represents
the base character and all of the combining characters, or it must
render the combination itself. Such rendering can be straight-forward,
but it is sometimes complicated when the combining marks interact with
each other, such as when there are two combining marks that would appear
above the same character. Formatting characters can also change the way
that a renderer would display text. Rendering can also be difficult
for some scripts that have complex display rules for base characters,
such as Arabic and Indic scripts.



. Standards Bodies and Standards

This section describes some of the standards bodies and standards that
appear in discussions of internationalization in the IETF. This is an
incomplete and possibly over-full list; listing too few bodies or
standards can be just as politically dangerous as listing too many. Note
that there are many other bodies that deal with internationalization;
however, few if any of them appear commonly in IETF standards work.

3.1 Standards bodies

ISO

The International Organization for Standardization has been involved
with standards for characters since before the IETF was started. ISO is
a non-governmental group made up of national bodies. ISO has many
diverse standards in the international characters area; the one that is
most used in the IETF is commonly referred to as "ISO/IEC 10646",
although its official name has more qualifications. (The IEC is
International Electrotechnical Commission). ISO/IEC 10646 describes a
CCS that covers almost all known written characters in use today.

ISO/IEC 10646 is controlled by the group known as "ISO/IEC JTC 1/SC 2
WG2", often called "WG2" for short. ISO standards go through many steps
before being finished, and years often go by between changes to ISO/IEC
. Information on WG2, and its work products, can be found at
<http://www.dkuug.dk/JTC1/SC2/WG2/>.

The standard, which comes in multiple parts, can be purchased in both
print and CD-ROM versions. One example of how to cite the standard is
given in [RFC2279]. Any standard that cites ISO/IEC 10646 needs to
evaluate how to handle the versioning problem that is relevant to the
protocol's needs.

ISO is responsible for other standards that might be of interest to
protocol developers. [ISO 639] specifies the names of languages, and
[ISO 3166] specifies the abbreviations of countries. Character work is
done in the group known as ISO/IEC JTC1/SC22 and ISO TC46, as well as
other ISO groups.

Another relevant ISO group is JTC 1/SC22/WG20, which is responsible for
internationalization in JTC1, such as for international string ordering.
Information on WG20, and its work products, can be found at
<http://www.dkuug.dk/jtc1/sc22/wg20/>

Unicode Consortium

The second important group for international character standards is the
Unicode Consortium. The Unicode Consortium is a trade association of
companies, governments, and other groups interested in promoting the
Unicode Standard [UNICODE]. The Unicode Standard is a CCS whose
repertoire and code points are identical to ISO/IEC 10646. The Unicode
Consortium has added features to the base CCS which make it more useful
in protocols, such as defining attributes for each character. Examples
of these attributes include case conversion and numeric properties.

The Unicode Consortium publishes addenda to the Unicode Standard as
Unicode Technical Reports. There are many types of technical reports at
various stages of maturity. The Unicode Standard and affiliated
technical reports can be found at <http://www.unicode.org/>.

World Wide Web Consortium (W3C)

This group created and maintains the standard for XML, the markup
language for text that has become very popular. XML has always been
fully internationalized so that there is no need for a new version to
handle international text.

local and regional standards organizations

Just as there are many native CCSs and charsets, there are many local
and regional standards organizations to create and support them. Common
examples of these are ANSI (United States), and CEN/ISSS (Europe).

3.2 Encodings and transformation formats of ISO/IEC 10646

Characters in the ISO/IEC 10646 CCS can be expressed in many ways.
Encoding forms are direct addressing methods, while transformation
formats are methods for expressing encoding forms as bits on the wire.

Basic Multilingual Plane (BMP)

The BMP is composed of the first 2^16 code points in ISO/IEC 10646. The
BMP is also called "plane 0".

UCS-2 and UCS-4

UCS-2 and UCS-4 are the two encoding forms defined for ISO/IEC 10646.
UCS-2 addresses only the BMP. Because many useful characters (such as
many Han characters) have been defined outside of the BMP, many people
would consider UCS-2 to be dead. Theoretically, UCS-4 addresses the
entire range of 2^31 code points from ISO/IEC 10646 as 32-bit values.
However, for interoperability with UTF-16, ISO 10646 restricts the range
of characters that will actually be allocated to the values 0..0x10FFFF.

UTF-8

UTF-8, a transformation format specified in [RFC2279], is the preferred
encoding for IETF protocols. Characters in the BMP are encoded as one,
two, or three octets. Characters outside the BMP are encoded as four
octets. Characters from the US-ASCII repertoire have the same
on-the-wire representation in UTF-8 as they do in US-ASCII.

UTF-16, UTF-16BE, and UTF-16LE

UTF-16, UTF-16BE, and UTF-16LE, three transformation formats defined in
[RFC2781], are not required by any IETF standards, and are thus used
much less often than UTF-8. Characters in the BMP are always encoded as
two octets, and characters outside the BMP are encoded as four octets.
The three formats differ based on the order of the octets and the
presence of a special lead-in mark called the "byte order mark" or
"BOM".

UTF-32

The Unicode Consortium has defined UTF-32 as a transformation format for
UCS-4 in [UTR19].

SCSU and BOCU-1

The Unicode Consortium has defined an encoding, SCSU, which is designed
to offer good compression for typical text. SCSU is described in [UTR6].
A different encoding that is meant to be MIME-friendly, BOCU-1, is
described in [UTN6]. Although compression is attractive, as opposed to
UTF-8 , neither of these (at the time of this writing) has attracted much
interest in the IETF.

3.3 Native CCSs and charsets

Before ISO/IEC 10646 was developed, many countries developed their own
CCSs and charsets. Many dozen of these are in common use on the Internet
today. Examples include ISO 8859-5 for Cyrillic and Shift-JIS for
Japanese scripts.

The official list of the registered charset names for use with IETF
protocols is maintained by IANA and can be found at
<http://www.iana.org/assignments/character-sets>. The list contains
preferred names and aliases. Note that this list has historically
contained many errors, such as names that are in fact not charsets
or references that do not give enough detail to reliably map names
to charsets.

Probably the most well-known native CCS is ASCII [US-ASCII]. This CCS is
used as the basis for keywords and parameter names in many IETF
protocols, and as the sole CCS in numerous IETF protocols that have not
yet been internationalized.

[UTR22] describes issues involved in mapping character data between
charsets, and an XML format for mapping table data.


. Character Issues

This section contains terms and topics that are commonly used in
character handling and therefore are of concern to people
internationalizing protocols. These topics are standardized outside the
IETF.

combining character

A member of an identified subset of the coded character set of ISO/IEC
 intended for combination with the preceding non-combining graphic
character, or with a sequence of combining characters preceded by a
non-combining character. <ISOIEC10646>

composite sequence

A sequence of graphic characters consisting of a non-combining character
followed by one or more combining characters. A graphic symbol for a
composite sequence generally consists of the combination of the graphic
symbols of each character in the sequence. A composite sequence is not a
character and therefore is not a member of the repertoire of ISO/IEC
. <ISOIEC10646>

In some CCSs, some characters consist of combinations of other
characters. For example, the letter "a with acute" might be a
combination of the two characters "a" and "combining acute", or it might
be a combination of the three characters "a", a non-destructive
backspace, and an acute. The rules for combining two or more characters
are called "composition rules", and the rules for taking apart a
character into other characters is called "decomposition rules". The
results of composition is called a "precomposed character"; the results
of decomposition is called a "decomposed character".

normalization

Normalization is the transformation of data to a normal form, for
example, to unify spelling. <UNICODE>

Note that the phrase "unify spelling" in the definition above does not
mean unifying different words with the same meaning (such as "color" and
"colour"). Instead, it means unifying different character sequences that
are intended to form the same composite characters (such as
"<a><n><combining tilde><o>" and "<a><n with tilde><o>").

The purpose of normalization is to allow two strings to be compared for
equivalence. The strings "<a><n><combining tilde><o>" and "<a><n with
tilde><o>" would be shown identically on a text display device. If a
protocol designer wants those two strings to be considered equivalent
during comparison, the protocol must define where normalization occurs.

The terms "normalization" and "canonicalization" are often used
interchangeably. Generally, they both mean to convert a string of one or
more characters into another string based on standardized rules. Some
CCSs allow multiple equivalent representations for a written string;
normalization selects one among multiple equivalent representations as a
base for reference purposes in comparing strings. In strings of
text, these rules are usually based on decomposing combined characters
or composing characters with combining characters. [UTR15] describes the
process and many forms of normalization in detail. Normalization is
important when comparing strings to see if they are the same.

case

Case is the feature of certain alphabets where the letters have two
distinct forms. These variants, which may differ markedly in shape and
size, are called the uppercase letter (also known as capital or
majuscule) and the lowercase letter (also known as small or minuscule).
Case mapping is the association of the uppercase and lowercase forms of
a letter. <UNICODE>

There is usually (but not always) a one-to-one mapping between the same
letter in the two cases. However, there are many examples of characters
which exist in one case but for which there is no corresponding
character in the other case or for which there is a special mapping
rule, such as the Turkish dotless "i" and some Greek characters with
modifiers. Case mapping can even be dependent on locale. Converting text
to have only one case is called "case folding".

sorting and collation

Collating is the process of ordering units of textual information.
Collation is usually specific to a particular language. It is sometimes known
as alphabetizing, although alphabetization is just a special case of
sorting and collation. <UNICODE>

Collation is concerned with the determination of the relative order of
any particular pair of strings, and algorithms concerned with collation
focus on the problem of providing appropriate weighted keys for string
values, to enable binary comparison of the key values to determine the
relative ordering of the strings.

Sorting is the process of actually putting data records into specified
orders, according to criteria for comparison between the records.
Sorting can apply to any kind of data (including textual data) for which
an ordering criterion can be defined. Algorithms concerned with sorting
focus on the problem of performance (in terms of time, memory, or other
resources) in actually putting the data records into a specified order.

A sorting algorithm for string data can be internationalized by
providing it with the appropriate collation-weighted keys corresponding
to the strings to be ordered.

Many processes have a need to order strings in a consistent sequence
(sorted). For only a few CCS/CES combinations, there is an obvious sort
order that can be done without reference to the linguistic meaning of
the characters: the codepoint order is sufficient for sorting. That is,
the codepoint order is also the order that a person would use in sorting
the characters. For most CCS/CES combinations, the codepoint order would
make no sense to a person and therefore is not useful for sorting if the
results will be displayed to a person.

Codepoint order is usually not how any human educated by a local school
system expects to see strings ordered; if one orders to the expectations
of a human, one has a language-specific sort. Sorting to codepoint order
will seem inconsistent if the strings are not normalized before sorting
because different representations of the same character will sort
differently. This problem may be smaller with a language-specific sort.

code table

A code table is a table showing the characters allocated to the octets
in a code. <ISOIEC10646>

Code tables are also commonly called "code charts".

4.1 Types of characters

The following definitions of types of characters do not clearly
delineate each character into one type, nor do they allow someone to
accurately predict what types would apply to a particular character. The
definitions are intended for application designers to help them think
about the many (sometimes confusing) properties of text.

alphabetic

An informative Unicode property. Characters that are the primary units
of alphabets and/or syllabaries, whether combining or noncombining. This
includes composite characters that are canonical equivalents to a
combining character sequence of an alphabetic base character plus one or
more combining characters: letter digraphs; contextual variant of
alphabetic characters; ligatures of alphabetic characters; contextual
variants of ligatures; modifier letters; letterlike symbols that are
compatibility equivalents of single alphabetic letters; and
miscellaneous letter elements. <UNICODE>

ideographic

Any symbol that primarily denotes an idea (or meaning) in contrast to a
sound (or pronunciation), for example, a symbol showing a telephone or
the Han characters used in Chinese, Japanese, and Korean. <UNICODE>

punctuation

Characters that separate units of text, such as sentences and phrases,
thus clarifying the meaning of the text. The use of punctuation marks is
not limited to prose; they are also used in mathematical and scientific
formulae, for example. <UNICODE>

symbol

One of a set of characters other than those used for letters, digits, or
punctuation, and representing various concepts generally not connected
to written language use per se. Examples include symbols for
mathematical operators, symbols for OCR, symbols for box-drawing or
graphics, and symbols for dingbats. <NONE>

Examples of symbols include characters for arrows, faces, and geometric
shapes. [UNICODE] has a property that defines characters as symbols.

nonspacing character

A combining character whose positioning in presentation is dependent on
its base character. It generally does not consume space along the visual
baseline in and of itself. <UNICODE>

A combining acute accent (U+0301) is an example of a nonspacing
character.

diacritic

A mark applied or attached to a symbol to create a new symbol that
represents a modified or new value. They can also be marks applied to a
symbol irrespective of whether it changes the value of that symbol. In
the latter case, the diacritic usually represents an independent value
(for example, an accent, tone, or some other linguistic information).
Also called diacritical mark or diacritical. <UNICODE>

control character

The 65 characters in the ranges U+0000..U+001F and U+007F..U+009F. They
are also known as control codes. <UNICODE>

formatting character

Characters that are inherently invisible but that have an effect on the
surrounding characters. <UNICODE>

Examples of formatting characters include characters for specifying the
direction of text and characters that specify how to join multiple
characters.

compatibility character

A graphic character included as a coded character of ISO/IEC 10646
primarily for compatibility with existing coded character sets.
<ISOIEC10646>

For example, U+FF01 (FULLWIDTH EXCLAMATION MARK) was included for
compatibility with Asian character sets that include full-width and
half-width ASCII characters.


. User interface for text

Although the IETF does not standardize user interfaces, many protocols
make assumptions about how a user will enter or see text that is used in
the protocol. Internationalization challenges assumptions about the type
and limitations of the input and output devices that may be used with
applications that use various protocols. It is therefore useful to
consider how users typically interact with text that might contain
one or more non-ASCII characters.

input methods

An input method is a mechanism for a person to enter text into an
application. <NONE>

Text can be entered into a computer in many ways. Keyboards are by far
the most common device used, but many characters cannot be entered on
typical computer keyboards in a single stroke. Many operating systems
come with system software that lets users input characters outside the
range of what is allowed by keyboards.

For example, there are dozens of different input methods for Han
characters in Chinese, Japanese, and Korean. Some start with phonetic
input through the keyboard, while others use the number of strokes in
the character. Input methods are also needed for scripts that have many
diacritics, such as European characters that have two or three
diacritics on a single alphabetic character.

rendering rules

A rendering rule is an algorithm that a system uses to decide how to
display a string of text. <NONE>

Some scripts can be directly displayed with fonts, where each character
from an input stream can simply be copied from a glyph system and put on
the screen or printed page. Other scripts need rules that are based on
the context of the characters in order to render text for display.

Some examples of these rendering rules include:

- Scripts such as Arabic (and many others), where the form of the letter
changes depending on the adjacent letters, whether the letter is
standing alone, at the beginning of a word, in the middle of a word, or
at the end of a word. The rendering rules must choose between two or
more glyphs.

- Scripts such as the Indic scripts, where consonants may change their
form if they are adjacent to certain other consonants or may be
displayed in an order different from the way they are stored and
pronounced. The rendering rules must choose between two or more
glyphs.

- Arabic and Hebrew scripts, where the order of the characters displayed
are changed by the bidirectional properties of the alphabetic characters
and with right-to-left and left-to-right ordering marks. The rendering
rules must choose the order that characters are displayed.

graphic symbol

A graphic symbol is the visual representation of a graphic character or
of a composite sequence. <ISOIEC10646>

font

A font is a collection of glyphs used for the visual depiction of
character data. A font is often associated with a set of parameters (for
example, size, posture, weight, and serifness), which, when set to
particular values, generate a collection of imagable glyphs. <UNICODE>

bidirectional display

The process or result of mixing left-to-right oriented text and
right-to-left oriented text in a single line is called bidirectional
display. <UNICODE>

Most of the world's written languages are displayed left-to-right.
However, many widely-used written languages such as ones based on the
Hebrew or Arabic scripts are displayed right-to-left. Right-to-left text
often confuses protocol writers because they have to keep thinking in
terms of the order of characters in a string in memory, and that order
might be different than what they see on the screen. (Note that some
languages are written both horizontally and vertically.)

Further, bidirectional text can cause confusion because there are
formatting characters in ISO/IEC 10646 which cause the order of display
of text to change. These explicit formatting characters change the
display regardless of the implicit left-to-right or right-to-left
properties of characters.

It is common to see strings with text in both directions, such as
strings that include both text and numbers, or strings that contain a
mixture of scripts.

[UNICODE] has a long and incredibly detailed algorithm for displaying
bidirectional text.

undisplayable character

A character that has no displayable form. <NONE>

For instance, the zero-width space (U+200B) cannot be displayed because
it takes up no horizontal space. Formatting characters such as those for
setting the direction of text are also undisplayable. Note, however,
that every character in [UNICODE] has a glyph associated with it, and
that the glyphs for undisplayable characters are enclosed in a
dashed square as an indication that the actual character is
undisplayable.


. Text in current IETF protocols

Many IETF protocols started off being fully internationalized, while
others have been internationalized as they were revised. In this
process, IETF members have seen patterns in the way that many protocols
use text. This section describes some specific protocol interactions
with text.

protocol elements

Protocol elements are uniquely-named parts of a protocol. <NONE>

Almost every protocol has named elements, such as "source port" in TCP.
In some protocols, the names of the elements (or text tokens for the
names) are transmitted within the protocol. For example, in SMTP and
numerous other IETF protocols, the names of the verbs are part of the
command stream.  The names are thus part of the protocol standard.
The names of protocol elements are not normally seen by end users.

name spaces

A name space is the set of valid names for a particular item, or the
syntactic rules for generating these valid names. <NONE>

Many items in Internet protocols use names to identify specific
instances or values. The names may be generated (by some prescribed
rules),  registered centrally (e.g., such as with IANA), or have a
distributed registration and control mechanism, such as the names in the
DNS.

on-the-wire encoding

The encoding and decoding used before and after transmission over the
network is often called the "on-the-wire" (or sometimes just "wire")
format. <NONE>

Characters are identified by codepoints. Before being transmitted in a
protocol, they must first be encoded as bits and octets. Similarly, when
characters are received in a transmission, they have been encoded, and a
protocol that needs to process the individual characters needs to decode
them before processing.

parsed text

Text strings that is analyzed for subparts. <NONE>

In some protocols, free text in text fields might be parsed. For
example, many mail user agents will parse the words in the text of the
Subject: field to attempt to thread based on what appears after the
"Re:" prefix.

charset identification

Specification of the charset used for a string of text. <NONE>

Protocols that allow more than one charset to be used in the same place
should require that the text be identified with the appropriate charset.
Without this identification, a program looking at the text cannot
definitively discern the charset of the text. Charset identification is
also called "charset tagging".

language identification

Specification of the human language used for a string of text. <NONE>

Some protocols (such as MIME and HTTP) allow text that is meant for
machine processing to be identified with the language used in the text.
Such identification is important for machine-processing of the text,
such as by systems that render the text by speaking it. Language
identification is also called "language tagging".

MIME

MIME (Multipurpose Internet Mail Extensions) is a message format that
allows for textual message bodies and headers in character sets other
than US-ASCII in formats that require ASCII (most notably, [RFC2822], the
standard for Internet mail headers). MIME is described in RFCs 2045
through 2049, as well as more recent RFCs. <NONE>

transfer encoding syntax

A transfer encoding syntax (TES) (sometimes called a transfer encoding
scheme) is a reversible transform of already-encoded data that is
represented in one or more character encoding schemes. <NONE>

TESs are useful for encoding types of character data into an another
format, usually for allowing new types of data to be transmitted over
legacy protocols. The main examples of TESs used in the IETF include
Base64 and quoted-printable.

Base64

Base64 is a transfer encoding syntax that allows binary data to be
represented by the ASCII characters A through Z, a through z, 0 through
9, +, /, and =. It is defined in [RFC2045]. <NONE>

quoted printable

Quoted printable is a transfer encoding syntax that allows strings that
have non-ASCII characters mixed in with mostly ASCII printable
characters to be somewhat human readable. It is described in [RFC2047].
<NONE>

The quoted printable syntax is generally considered to be a
failure at being readable. It is jokingly referred to as "quoted
unreadable".

XML

XML (which is an approximate abbreviation for Extensible Markup
Language) is a popular method for structuring text. XML text is
explicitly tagged with charsets. The specification for XML can be found
at <http://www.w3.org/XML/>. <NONE>

ASN.1 text formats

The ASN.1 data description language has many formats for text data. The
formats allow for different repertoires and different encodings. Some of
the formats that appear in IETF standards based on ASN.1 include
IA5String (all ASCII characters), PrintableString (most ASCII
characters, but missing many punctuation characters), BMPString
(characters from ISO/IEC 10646 plane 0 in UTF-16BE format), UTF8String
(just as the name implies), and TeletexString (also called T61String;
the repertoire changes over time).

ASCII-compatible encoding (ACE)

Starting in 1996, many ASCII-compatible encoding schemes (which are
actually transfer encoding syntaxes) have been proposed as possible
solutions for internationalizing host names. Their goal is to be able to
encode any string of ISO/IEC 10646 characters as legal DNS host names
(as described in STD 13). At the time of this writing, no ACE has become
an IETF standard.


. Other Common Terms In Internationalization

This is a hodge-podge of other terms that have appeared in
internationalization discussions in the IETF. It is likely that
additional terms will be added as this document matures.

locale

Locale is the user-specific location and cultural information
managed by a computer.  <NONE>

Because languages differ from country to country (and even region to
region within a country), the locale of the user can often be an
important factor. Typically, the locale information for a user includes
the language(s) used.

Locale issues go beyond character use, and can include things such as
the display format for currency, dates, and times. Some locales
(especially the popular "C" and "POSIX" locales) do not include language
information.

It should be noted that there are many thorny, unsolved issues with
locale. For example, should text be viewed using the locale information
of the person who wrote the text or the person viewing it? What if the
person viewing it is travelling to different locations? Should only
some of the locale information affect creation and editing of text?

Latin characters

"Latin characters" is a not-precise term for characters historically
related to ancient Greek script and currently used throughout the world.
<NONE>

The base Latin characters make up the ASCII repertoire and have been
augmented by many single and multiple diacritics and quite a few other
characters. ISO/IEC 10646 encodes the Latin characters in the ranges
U+0020..U+024F, U+1E00..U+1EFF, and other ranges.

romanization

The transliteration of a non-Latin script into Latin characters. <NONE>

Because of the widespread use of Latin characters, people have tried to
represent many languages that are not based on a Latin repertoire in
Latin. For example, there are two popular romanizations of Chinese:
Wade-Giles and Pinyin, the latter of which is by far more common today.
Most romanization systems are inexact and do not give perfect round trip
mappings between the native script and the Latin characters.

CJK characters and Han characters

The ideographic characters used in Chinese, Japanese, Korean, and
traditional Vietnamese writing systems are often called 'CJK characters'
after the initial letters of the language names in English. They are
also called "Han characters", after the term in Chinese that is often
used for these characters. <NONE>

Note that CJK and Han characters do not include the phonetic characters
used in the Japanese and Korean languages.

In ISO/IEC 10646, the Han characters were "unified", meaning that each
set of Han characters from Japanese, Chinese, and/or Korean that had the
same origin was assigned a single code point. The positive result of
this was that many fewer code points were needed to represent Han; the
negative result of this was that characters that people who write the
three languages think are different have the same code point. There is a
great deal of disagreement on the nature, the origin, and the severity
of the problems caused by Han unification.

translation

The process of conveying the meaning of some passage of text in one
language, so that it can be expressed equivalently in another language.
<NONE>

Most language translation systems are inexact and cannot be applied
repeatedly to go from one language to another to another.

transliteration

The process of representing the characters of an alphabetical or
syllabic system of writing by the characters of a conversion alphabet.
<NONE>

Most script transliterations are exact, and many have perfect round-trip
mappings. The notable exception to this is romanization, described
above. Transliteration involves converting text expressed in one script
into another script, generally on a letter-by-letter basis.

transcription

The process of systematically writing the sounds of some passage of
spoken language, generally with the use of a technical phonetic alphabet
(usually Latin-based) or other systematic transcriptional orthography.
Transcription also sometimes refers to the conversion of written text
into a transcribed (usually Latin-based) form, based on the sound of the
text as if it had been spoken. <NONE>

Unlike transliterations, which are generally designed to be round-trip
convertible, transcriptions of written material are almost never
round-trip convertible to their original form.

regular expressions

Regular expressions provide a mechanism to select specific strings from
a set of character strings. Regular expressions are a language used to
search for text within strings, and possibly modify the text found with
other text. <NONE>

Pattern matching for text involves being able to represent one or more
code points in an abstract notation, such as searching for all capital
Latin letters or all punctuation. The most common mechanism in IETF
protocols for naming such patterns is the use of regular expressions.
There is no single regular expression language, but there are numerous
very similar dialects.

The Unicode Consortium has a good discussion about how to adapt regular
expression engines to use Unicode. [UTR18]

private use

ISO/IEC 10646 code points from U+E000 to U+F8FF, U+F0000 to U+FFFFD, and
U+100000 to U+10FFFD are available for private use. This refers to code
points of the standard whose interpretation is not specified by the
standard and whose use may be determined by private agreement among
cooperating users. <UNICODE>

The use of these "private use" characters is defined by the parties who
transmit and receive them, and is thus not appropriate for
standardization. (The IETF has a long history of private use names for
things such as "x-" names in MIME types, charsets, and languages. The
experience with these has been quite negative, with many implementors
assuming that private use names are in fact public and long-lived.)


. Security Considerations

Security is not discussed in this document.


. References

[CHARMOD] Character Model for the World Wide Web 1.0, W3C,
<http://www.w3.org/TR/charmod/>

[FRAMEWORK] ISO/IEC TR 11017:1997(E). Information technology - Framework
for internationalization, prepared by ISO/IEC JTC 1/SC 22/WG 20.

[ISO 639]   ISO 639:2000 (E/F) - Code for the representation of names of
languages.

[ISO 3166]  ISO 3166:1988 (E/F) - Codes for the representation of names
of countries.

[ISOIEC10646] ISO/IEC 10646-1:2000. International Standard --
Information technology -- Universal Multiple-Octet Coded Character Set
(UCS) -- Part 1: Architecture and Basic Multilingual Plane.

[RFC2045] "MIME Part One: Format of Internet Message Bodies", RFC 2045,
N. Freed and N. Borenstein.

[RFC2047] "MIME Part Three: Message Header Extensions for Non-ASCII
Text", RFC 2047, K. Moore.

[RFC2277] "IETF Policy on Character Sets and Languages", RFC 2277, H.
Alvestrand.

[RFC2279] "UTF-8, a transformation format of ISO 10646", RFC 2279, F.
Yergeau.

[RFC2781] "UTF-16, an encoding of ISO 10646", RFC 2781, P. Hoffman and
F. Yergeau.

[RFC2822] "Internet Message Format", RFC 2822, P. Resnick.

[RFC3066] "Tags for the Identification of Languages", RFC 3066, H.
Alvestrand.

[UNICODE] The Unicode Consortium, "The Unicode Standard -- Version 3.0",
ISBN 0-201-61633-5. Described at
<http://www.unicode.org/unicode/standard/versions/Unicode3.0.html>.

[US-ASCII]  Coded Character Set -- 7-bit American Standard Code for
Information Interchange, ANSI X3.4-1986.

[UTN6] "BOCU-1: MIME-Compatible Unicode Compression", Unicode Technical
Note #6, M. Scherer & M. Davis.

[UTR6] "A Standard Compression Scheme for Unicode", Unicode Technical
Report #6, M. Wolf, et. al.

[UTR15] "Unicode Normalization Forms", Unicode Technical Report #15, M.
Davis & M. Duerst.

[UTR18] "Unicode Regular Expression Guidelines", Unicode Technical
Report #18, M. Davis.

[UTR19] "UTF-32", Unicode Technical Report #19, M. Davis.

[UTR22] "Character Mapping Markup Language", Unicode Technical Report
#22, M. Davis.


. Additional Interesting Reading

ALA-LC Romanization Tables, Randall Barry (ed.), U.S. Library
of Congress, 1997, ISBN 0844409405

Blackwell Encyclopedia of Writing Systems, Florian Coulmas, Blackwell
Publishers, 1999, ISBN 063121481X

The World's Writing Systems, Peter Daniels and William Bright, Oxford
University Press, 1996, ISBN 0195079930

Writing Systems of the World, Akira Nakanishi, Charles E. Tuttle
Company, 1980, ISBN 0804816549


. Index

alphabetic -- 4.1
ASCII-compatible encoding (ACE) -- 6
ASN.1 text formats -- 6
Base64 -- 6
Basic Multilingual Plane (BMP) -- 3.2
bidirectional display -- 5
BOCU-1 -- 3.2
case -- 4
character -- 2
character encoding form -- 2
character encoding scheme -- 2
charset -- 2
charset identification -- 6
CJK characters and Han characters -- 7
code chart -- 4
code table -- 4
coded character -- 2
coded character set -- 2
combining character -- 4
compatibility character -- 4.1
composite sequence -- 4
control character -- 4.1
diacritic -- 4.1
displaying and rendering text -- 2
font -- 5
formatting character -- 4.1
glyph -- 2
glyph code -- 2
graphic symbol -- 5
i18n, l10n -- 2
ideographic -- 4.1
input methods -- 5
internationalization -- 2
ISO -- 3.1
language -- 2
language identification -- 6
Latin characters -- 7
local and regional standards organizations -- 3.1
locale -- 7
localization -- 2
MIME -- 6
multilingual -- 2
name spaces -- 6
nonspacing character -- 4.1
normalization -- 4
on-the-wire encoding -- 6
parsed text -- 6
private use -- 7
protocol elements -- 6
punctuation -- 4.1
quoted printable -- 6
regular expressions -- 7
rendering rules -- 5
romanization -- 7
script -- 2
SCSU -- 3.2
sorting and collation -- 4
symbol -- 4.1
transcoding -- 2
transcription -- 7
transfer encoding syntax -- 6
translation -- 7
transliteration -- 7
UCS-2 and UCS-4 -- 3.2
undisplayable character -- 5
Unicode Consortium -- 3.1
UTF-32 -- 3.2
UTF-16, UTF-16BE, and UTF-16LE -- 3.2
UTF-8 -- 3.2
World Wide Web Consortium -- 3.1
XML -- 6


A. Acknowledgements

The definitions in this document come from many sources, including a
wide variety of IETF documents.

James Seng contributed to the initial outline of this document. Harald
Alvestrand and Martin Duerst made extensive useful comments on early
versions. Others who contributed to the development include:

Dan Kohn
Jacob Palme
Johan van Wingen
Peter Constable
Yuri Demchenko
Susan Harris
Zita Wenzel
John Klensin
Henning Schulzrinne
Leslie Daigle
Markus Scherer
Ken Whistler


B. Author Contact Information

Paul Hoffman
Internet Mail Consortium and VPN Consortium
 Segre Place
Santa Cruz, CA  95060 USA
paul.hoffman@imc.org and paul.hoffman@vpnc.org


C. Changes from -07 to -08

[[ To be removed when published as an RFC ]]

Small spelling and punctuation corrections. Also fixed some ambiguous
references.

Changed "internationalized text" to other more appropriate words in
many parts of the text.

2, removed the reference to the IDN requirements document because it
doesn't exist.

2, character: Added second sub-definition and changed the wording of the
following paragraph.

2, displaying and rendering text: Corrected last sentence of second
paragraph by removing "do not use combining characters but".

3.1, ISO: Added SC22/WG20. Also changed the URL for SC2.

4, case: Added "or for which there is a special mapping rule".

4, code table: Added alias to "code chart".

4, sorting and collation: Added 2nd, 3rd, and 4th paragraphs to make the
difference clearer. Made it clearer that codepoint order is rarely the
best one. Also changed "localized" to "language-specific".

4.1, symbol: Completely changed definition.

4.1, nonspacing character: Completely changed definition and example.

4.1, formatting character: Removed "letter-spacing characters".

5, bidirectional display: Rewrote third paragraph.

7, CJK and Han: Reworded the first sentence. Added the last
sentence.

. transliteration: changed the definition and added the last sentence.

7, transcription: added this definition.

7, regular expressions: Added discussion of UTR 18.

9: Added reference to UTR 18.

A. Added Ken Whistler