Proposal
Extension of CSS numbering systems

Author : Daniel Glazman, Electricité de France, Research and Development Division
Date : 18 june1998

Introduction

CSS 2 introduces a set of numbering systems for lists (list-style-type property) and counters (counter() and counters() functional notations). Counters functional notations can use a value of the list-style-type property that will rule the rendition. Here is the current CSS 2 list-style-property definition :

This definition does not cover all document providers' needs and is restrictive in comparison with common text editors (e.g. Microsoft Word). It has been proposed, during the May 1998 ftf meeting in Clamart to extend in CSS 3 the list of handled numbering systems :

• aphabets using a traditional numbering system
• cardinals (1st, 2nd, 3rd, ...)
• word (one, two, three, ...)
• gender (in French for instance : Premier chapitre, Second Chapitre, ... but Première partie, Seconde partie, ...)
• ...

Here is a non exhaustive list of the most common traditional numbering systems of the world (except western-decimal). Armenian and Gorgian are not in this list because I've already posted a long message about these important and specific numbering systems.

Where and how ?

The list of values above is clearly mixing a lot of notions :

• numbering system (including no numbering : circle, square and disc)
• style : numbers base 10, alphabetical, multiplicative-additive
• case
• leading number

Extension of CSS numbering systems to other alphabets (ethiopic a.k.a. amharic, malayalam, cinghalese, ...) and to other styles (ordinals, cardinals, ordinal words, cardinal words) raises some problems :

1. extension of the list of values above to new alphabets is possible but introduction of ordinals, cardinals and words will dramatically increase the length of that list; furthermore, it is not the common practice in CSS to agregate different notions in one single namespace.
2. other possible solutions must preserve the counter() and counters() functional notations.

Because of CSS error handling (section 4.2), it is not possible to add a second value to list-style-type property and preserve a backwards compatibility. On an other hand, new browsers should preserve correct rendering for CSS 2 conformant style sheets.

Proposal

It is proposed to add new properties for list numbering and turn list-style-type property into a shorthand property. This will allow backwards compatibility : CSS 2 style sheets using list markers will also be CSS 3 style sheets. Some of the properties below are allowed with another media than visual : aural, for accessibility reasons.

This property specifies appearance of the list item marker.

none

No marker specified

symbol

if 'list-style-image' has the value none or if the image pointed to by the URI cannot be displayed, appearance of the marker is specified by the value of 'list-numbering-type' property.

cardinals

marker is a cardinal number, the numbering system being given by the value of 'list-numbering-type' property. For instance in English : 1, 2, 3, 4, ... or one, two, three, four, ...

ordinals

marker is an ordinal number, the numbering system being given by the value of 'list-numbering-type' property. For instance in English : 1st, 2nd, 3rd, 4th, ... or first, second, third, forth, ...

This property specifies the aspect of the marker if 'list-numbering-style' has the value ordinals or cardinals.

figures

numbering must be done using numbers, the numbering system being given by the value of 'list-numbering-type' property.

words

numbering must be done using the word equivalent of each number, the numbering system being given by the value of 'list-numbering-type' property. The language of the list item element is used to determine this word.

Values in italic are here for compatibility reasons with CSS 2 and are deprecated. The -decimal, -alphabetical or -traditional version should be used instead. Case should be indicated using the 'list-style-type' shorthand property. The ethiopic value becomes 'amharic-traditional' since Ethiopic is the name of a language and Amharic is the name of the associated script, used by Ethiopic and three other languages from Ethiopia and Eryrthrea. All values implying numbering based on ascii letters are deprecated in favor of western-alphabetical.

Glyphs markers :

disc

square

circle

glyph markers if 'list-numbering-style' has value symbol

Decimal numbering systems :

decimal

western-decimal

decimal numbers, beginning with 1

leading-zero

Decimal numbers 01, 02, 03 ... 99 (or 001, 002, 003, ... 999 if more than 99 items) ...

arabic-decimal

Arabic-Indic decimal numbers, beginning with 1

eastern-arabic-decimal

Eastern Arabic-Indic decimal numbers for Persian and Urdu, beginning with 1

devanagari-decimal

bengali-decimal

gurmukhi-decimal

gujarati-decimal

orija-decimal

tamil-decimal

telugu-decimal

kannada-decimal

malayalam-decimal

thai-decimal

lao-decimal

tibetan-decimal

singhalese-decimal

khmer-decimal

dhivahi-decimal

mongolian-decimal

burmese-decimal

Decimal numberings for Devanagari, Bengali, Gurmukhi, Gujarati, Orija, Tamil, Telugu, Kannada, Malayalam, Thaï, Lao, Tibetan, Singhalese, Khmer, Dhivahi (Maldivian), Mongolian and Burmese scripts, beginning with 1.

Alphabetical numbering systems :

greek

greek-alphabetical

lower-greek

lower case greek numerals (alpha, beta, gamma, ...)

latin

western-alphabetical

lower-latin

alpha

lower-alpha

Lower case ascii letters (a, b, c, ... z). Rendering is undefined after 26 items.

upper-latin

upper-alpha

Upper case ascii letters (A, B, C, ..., Z). Rendering is undefined after 26 items.

Traditional numbering systems :

roman

roman-traditional

lower-roman

lower case roman numerals (i, ii, iii, iv, v, ...)

upper-roman

upper case roman numerals (I, II, III, IV, V, ...)

armenian

armenian-traditional

Traditional lower case Armenian additive numbering (ayb, ben, gim, ..., zhe, zhe-ayb, ...)

georgian

georgian-traditional

Traditional Georgian (Mkhedruli script) additive numbering (an, ban, gan, ..., he, tan, in, in-an, ...)

hebrew

hebrew-traditional

Traditional Hebrew additive numbering

cjk-ideographic

cjk-ideographic-traditional

Plain Chinese ideographic numbers

hiragana

hiragana-traditional

Modern Japanese hiragana numbering according to a syllable table (a, i, u, e, o, ka, ki, ...)

katakana

katakana-traditional

Modern Japanese katakana numbering according to a syllable table (A, I, U, E, O, KA, KI, ...)

hiragana-iroha

hiragana-iroha-traditional

Traditional Japanese hiragana numbering according to a famous poem (i, ro, ha, ni, ho, he, to, ...)

katakana-iroha

katakana-iroha-traditional

Traditional Japanese katakana numbering according to a famous poem (I, RO, HA, NI, HO, HE, TP, ...)

tamil-traditional

malayalam-traditional

singhalese-traditional

Traditional multiplicative-additive numbering systems for Tamil, Malayalam and Singhalese

ethiopic

amharic-traditional

Traditional Amharic multiplicative-additive numbering

This is now a shorthand property. The none initial value of 'test-transform' implies a lower case numbering if this information is relevant.

Leading zero

Without implicit evidence that other decimal systems than western-decimal use the leading zero notation, it is suggested to drop this problem and not to build a new property.

Counters

The definition of counter() functional notation remains unchanged. It can still use as second argument the values of 'list-style-type' shorthand property.

Gender and other grammatical relations

It is extremely difficult to obtain the gender of each marker of a numbering sequence according to the text before or after the marker and the language of this text. In some language, for instance from the finno-hungarian group, numbers can also be accorded with the declination ( ?) of the following word. In Estonian and Finnish, there are at least 12 declinations per word !...

Even if I expressed a very strong interest in this feature during the Clamart ftf meeting, I suggest to drop this problem or follow ChrisW proposal and discuss a way to declare that the user agent should, if it can handle such an operation, try to modify the gender, number and declination of the list marker depending on the contents of the containing LI and leave that user-agent-dependant. I see no other way to solve this problem in simple terms that we could add to CSS.

As Martin Dürst pointed it out in the w3c-css-wg mailing list, the gender information is extremely important for ACSS rendering of a list marker using a numbering system. I am very circumspect about the way to set this information. Is it an attribute on the list item element or is it a style of the marker only ? What happens if the contents of the list item - and its gender - is modified ? If the gender of the marker is handled by CSS, fixing the gender needs a second operation. As a document provider, I don't think this is normal.

In my opinion, this problem may be linked with content and I18N, maybe not with styles. Are a namespace of grammatical attributes and selection on these attributes in CSS a correct answer ? (naive question)

Annex 1 : Numbering systems

amharic-traditional

arabic-decimal

armenian-traditional (lower case on 1st line, upper case on 2nd)

bengali-decimal

burmese-decimal

cjk-ideographic-traditional

devanagari-decimal

dhivehi-decimal (Maldivian)

eastern-arabic-decimal

georgian-traditional

gujarati-decimal

gurmukhi-decimal

hebrew-traditional

khmer-decimal

lao-decimal

roman-traditional

malayalam-decimal

malayalam-traditional

mongolian-decimal

orija-decimal

singhalese-traditional (-decimal : from 1 to 9 only)

tamil-traditional (-decimal : from 1 to 9 only)

telugu-decimal

thai-decimal

tibetan-decimal