Unicode Versions and Unicode Technical Reports
The Unicode standard includes a group of supplementary documents known as Unicode Technical Reports. A snapshot of these is also included on the CD that comes with the book The Unicode Standard, Version 3.0, but the CD accompanying the Unicode 3.0 standard is now so out of date as to be nearly useless. You can always find the most up-to-date slate of technical reports on the Unicode Web site (http://www.unicode.org).
There are three kinds of technical reports:
Unicode Standard Annexes (UAX) are actual addenda and amendments to the Unicode standard.
Unicode Technical Standards (UTS) are adjunct standards related to Unicode. They're not normative parts of the standard itself, but carry their own conformance criteria.
Unicode Technical Reports (UTR) include various types of adjunct information, such as text clarifying or expanding on parts of the standard, implementation guidelines, and descriptions of procedures for doing things with Unicode text that don't rise to the level of official Unicode Technical Standards.
Other types of reports exist as well:
Draft Unicode Technical Reports (DUTR). Technical reports are often published while they're still in draft form. DUTR status indicates that an agreement has been reached in principle to adopt the proposal, but details must still be worked out. Draft technical reports don't have any normative force until the Unicode Technical Committee votes to remove "Draft" from their name, but they're published early to solicit comments and give implementers a head start.
Proposed Draft Unicode Technical Reports (PDUTR). These technical reports are published before an agreement in principle has been reached to adopt them.
The status of the technical reports is constantly changing. Here's a summary of the slate of technical reports as of this writing (January 2002).
Unicode Standard Annexes
All of the following Unicode Standard Annexes are officially part of the Unicode 3.2 standard:
UAX #9: The Unicode Bidirectional Algorithm. This report specifies the algorithm for laying out lines of text that mix left-to-right characters with right-to-left characters. It supersedes the description of the bidirectional layout algorithm in the Unicode 3.0 book. For an overview of the bidirectional layout algorithm, see Chapter 8. For implementation details, see Chapter 16.
UAX #11: East Asian Width. It specifies a set of character properties that determine how many display cells a character takes up when used in the context of East Asian typography. For more information, see Chapter 10.
UAX #13: Unicode Newline Guidelines. There are some interesting issues regarding how you represent the end of a line or paragraph in Unicode text, and this document clarifies them. This information is covered in Chapter 12.
UAX #14: Line Breaking Properties. This document specifies how a word-wrapping routine should treat the various Unicode characters. Word-wrapping is covered in Chapter 16.
UAX #15: Unicode Normalization Forms. Because Unicode has multiple ways of representing a lot of characters, it's often helpful to convert Unicode text to some kind of normalized form that prefers one representation for any given character over all the others. There are four Unicode Normalization Forms, described in this document. The Unicode Normalization Forms are covered in Chapter 4.
UAX #19: UTF-32. It specifies the UTF-32 encoding form. UTF-32 is covered in Chapter 6.
UAX #27: Unicode 3.1. This official definition includes the changes and additions to Unicode 3.0 that form Unicode 3.1. This document officially incorporates the other Unicode Standard Annexes into the standard and gives them a version number. It also includes code charts and character lists for all new characters added to the standard in Unicode 3.1.
Unicode Technical Standards
UTS #6: A Standard Compression Scheme for Unicode. It defines SCSU, a character encoding scheme for Unicode that results in serialized Unicode text of comparable size to the same text in legacy encodings. For more information, see Chapter 6.
UTS #10: The Unicode Collation Algorithm. It specifies a method of comparing character strings in Unicode in a language-sensitive manner and a default ordering of all characters to be used in the absence of a language-specific ordering. For more information, see Chapter 15.
Unicode Technical Reports
UTR #16: UTF-EBCDIC. It specifies a special 8-bit transformation of Unicode for use on EBCDIC-based systems. This topic is covered in Chapter 6.
UTR #17: Character Encoding Model. It defines a set of useful terms for discussing the various aspects of character encodings. This material was covered in Chapter 2 and reiterated in this chapter.
UTR #18: Regular Expression Guidelines. It provides some guidelines for how a regular-expression facility should behave when operating on Unicode text. This material is covered in Chapter 15.
UTR #20: Unicode in XML and Other Markup Languages. It specifies some guidelines for how Unicode should be used in the context of a markup language such as HTML or XML. This topic is covered in Chapter 17.
UTR #21: Case Mappings. It gives more detail than the standard itself on the process of mapping uppercase to lowercase, and vice versa. Case mapping is covered in Chapter 14.
UTR #22: Character Mapping Tables. It specifies an XML-based file format for describing a mapping between Unicode and some other encoding standard. Mapping between Unicode and other encodings is covered in Chapter 14.
UTR #24: Script Names. It defines an informative character property: the script name, which identifies the writing system (or "script") to which each character belongs. The script name property is discussed in Chapter 5.
Draft and Proposed Draft Technical Reports
PDUTR #25: Unicode Support for Mathematics. It gives a detailed account of the various considerations involved in using Unicode to represent mathematical expressions, including issues such as spacing and layout, font design, and interpretation of Unicode characters in the context of mathematical expressions. This technical report also includes a heuristic for detecting mathematical expressions in plain Unicode text, and proposes a scheme for representing structured mathematical expressions in plain text. We'll look at math symbols in Chapter 12.
DUTR #26: Compatibility Encoding Scheme for UTF-16: 8-bit (CESU-8). It documents a UTF-8-like Unicode encoding scheme that's being used in some existing systems. We'll look at CESU-8 in Chapter 6.
PDUTR #28: Unicode 3.2. Together with UAX #27, this official definition gives all the changes and additions to Unicode 3.0 that define Unicode 3.2, including revised and updated code charts with the new Unicode 3.2 characters. Unicode 3.2 is scheduled to be released in March 2002, so this technical report will likely be UAX #28 by the time you read these words.
Superseded Technical Reports
The following technical reports have been superseded by (or absorbed into) more recent versions of the standard:
UTR #1: Myanmar, Khmer, and Ethiopic. Absorbed into Unicode 3.0.
UTR #2: Sinhala, Mongolian, and Tibetan. Tibetan was in Unicode 2.0; Sinhala and Mongolian were added in Unicode 3.0.
UTR #3: Various Less-Common Scripts. This document includes exploratory proposals for historical or rare writing systems. It has been superseded by more recent proposals. Some of the scripts in this proposal have been incorporated into more recent versions of Unicode: Cherokee (Unicode 3.0), Old Italic (Etruscan; Unicode 3.1), Thaana (Maldivian; 3.0), Ogham (3.0), Runic (3.0), Syriac (3.0), Tagalog (3.2), Buhid (3.2), and Tagbanwa (3.2). Most of the others are in various stages of discussion, with another batch scheduled for inclusion in Unicode 4.0. This document is mostly interesting as a list of writing systems that will probably be in future versions of Unicode. For more information on this subject, check out http://www.unicode.org/unicode/alloc/Pipeline.html.
UTR #4: Unicode 1.1. Superseded by later versions.
UTR #5: Handling Non-spacing Marks. Incorporated into Unicode 2.0.
UTR #7: Plane 14 Characters for Language Tags. Incorporated into Unicode 3.1.
UTR #8: Unicode 2.1. Superseded by later versions.
The missing numbers belong to technical reports that have been withdrawn by their proposers, have been turned down by the Unicode Technical Committee, or haven't been published yet.
Many of the technical reports either define certain versions of Unicode or are superseded by certain versions of Unicode. Each version of Unicode comprises a particular set of characters, a particular version of the character-property files, and a certain set of rules for dealing with them. All of these things change over time (although certain things are guaranteed to remain the samesee "Unicode Stability Policies" later in this chapter).
This point brings us to the question of Unicode version numbers. A Unicode version number consists of three partsfor example, "Unicode 2.1.8." The first number is the major version number. It gets bumped every time a new edition of the book is released. That happens when the accumulation of technical reports becomes too unwieldy or when a great many significant changes (such as lots of new characters) are incorporated into the standard at once. A new major version of Unicode appears every several years.
The minor version number gets bumped whenever new characters are added to the standard or other significant changes are made. New minor versions of Unicode don't get published as books, but do get published as Unicode Standard Annexes. There has been one minor version of Unicode between each pair of major versions (i.e., there was a Unicode 1.1 [published as UTR #4], a Unicode 2.1 [UTR #8], and a Unicode 3.1 [UAX #27]), but this pattern was broken with the release of Unicode 3.2 (PDUTR #28 at the time of this writing, but most likely UAX #28 by the time you read this).
The update version number gets bumped when changes are made to the Unicode Character Database. Updates are published as new versions of the database; there is no corresponding technical report.
The current version of Unicode at the time of this writing (January 2002) is Unicode 3.1.1. Unicode 3.2, which includes a whole slate of new characters, is currently in beta and will likely be the current version by the time this book is published.
One has to be a bit careful when referring to a particular version of the Unicode standard from another document, particularly another standard. Unicode changes constantly, so it's seldom a good idea to nail yourself to one specific version. It's generally best either to specify only a major version number (or, in some cases, just major and minor version numbers) or to specify an open-ended range of versions (e.g., "Unicode 2.1 or later"). Generally, this approach is okay, as future versions of Unicode will only add charactersbecause you're never required to support a particular character, you can pin yourself to an open-ended range of Unicode versions and not sweat the new characters. Sometimes, however, changes are made to a character's properties in the Unicode Character Database that could alter program behavior. Usually, this is a good thingchanges are made to the database when the database is deemed to have been wrong beforebut your software may need to deal with this modification in some way.
Unicode Stability Policies
Unicode will, of course, continue to evolve. Nevertheless, you can count on some things to remain stable:
Characters that are in the current standard will never be removed from future standards. They may, in unusual circumstances, be deprecated (i.e., their use might be discouraged), but they'll never be eliminated and their code point values will never be reused to refer to different characters.
Characters will never be reassigned from one code point to another. If a character has ambiguous semantics, a new character may be introduced with more specific semantics, but the old one will never be taken away and will continue to have ambiguous semantics.
Character names will never change. Occasionally a character with ambiguous semantics will get out of sync with its name as its semantics evolve, but this is very rare.
Text in one of the Unicode Normalized Forms will always be in that normalized form. That is, the definition of the Unicode Normalized Forms will not change between versions of the standard in ways that would ª cause text that is normalized according to one version of the standard not to be normalized in later versions of the standard.
A character's combining class and canonical and compatibility decompositions will never change, as that would break the normalization guarantee.
A character's properties may change, but not in a way that would alter the character's fundamental identity. In other words, the representative glyph for "A" won't change to "B", and the category for "A" won't change to "lowercase letter." You'll see property changes only to correct clear mistakes in previous versions.
Various structural aspects of the Unicode character properties will remain the same, as implementations depend on some of these things. For example, the standard won't add any new general categories or any new bi-di categories, it will keep characters' combining classes in the range from 0 to 255, noncombining characters will always have a combining class of 0, and so on.
You can generally count on characters' other normative properties not changing, although the Unicode Consortium certainly reserves the right to fix mistakes in these properties.
The Unicode Consortium can change a character's informative properties more or less at will, without changing version numbers, because you don't have to follow them anyway. Again, this type of change shouldn't happen much, except when consortium members need to correct a mistake of some kind.
These stability guarantees are borne out of bitter experience. Characters did get removed and reassigned and characters' names did change in Unicode 1.1 as a result of the merger with ISO 10646, and it caused serious grief. This problem won't happen again.