Re: Vertical BIDI

From: Philippe Verdy (
Date: Tue May 18 2004 - 18:05:40 CDT

  • Next message: fantasai: "Re: Vertical BIDI"

    From: <>
    > Andrew C. West scripsit:
    > > It does ? I thought that the whole point of much of the recent discussion
    > > the uncertainty of how Ogham should be laid out in vertically formatted
    > > such as when embedded in Mongolian or vertical Chinese.
    > What's uncertain is whether a lr or a rl progression is favored, given the
    > paucity of evidence. Michael favors lr progression. There is no question
    > that the text is read BTT.

    This creates an interesting problem: Put in the same sentence Han (Chinese) and
    Mongolian words in a vertical layout (I don't think this is unlikely, as
    Mongolian is also spoken in China, and there's also a Chinese community in
    Mongolia). So Chinese ideographs will be laid out vertically from top to bottom
    (but not rotated, except for a few characters like ideographic punctuation marks
    or symbols), and Mongolian will be laid out from bottom to top in their normal
    stack orientation. Such a text is clearly bidirectional, so we would need BiDi
    processing to order glyphs correctly.

    Now admit that you want to present it horizontally: Han ideographs will not be
    rotated but will flow on rows from left to right. Suppose you have performed the
    Bidi processing according to the previous vertical presentation, then Mongolian
    stacks will flow from right to left (but unlike Han ideographs, they will need
    to be rotated...)

    Now try including some Latin words in this text (also not unlikely: there are
    lots of trademarks and people names that will need to be written with their
    normal Latin characters). If the text is presented vertically, there's a
    legitimate question of whever Latin should be rotated (but it will keep the Han
    flow direction.); according to Japanese practices with halfwidth characters (the
    standard Latin script is "half-width"), it will be rotated exactly like for
    "half-width" Hiragana or Katakana, simply because "half-width" characters align
    more naturally in that direction, and because the perception of character's
    baseline really requires (for standard presentations of texts) that baselines
    not be splitted at each character like in vertical columns of crosswords. This
    would then flow Latin in the same direction as Han, but the reverse direction of

    But if experience shows that Mongolian would be presented horizontally with lr
    direction, like with Latin, there's a problem.

    So what is shown here is that Bidi properties are only accurate for horizontal
    flows of text. What is missing is a separate set of Bidi properties for the
    vertical direction of the same flow... We could define basically a similar
    algorithm for vertical BiDi, but this would also require new BiDi properties.
    May be this would not require changing the existing Bidi properties for
    characters that are normally presented horizontally, but new property _values_
    would be needed for those that flow normally vertically(Ogham, Mongolian)...
    Plus a mapping of these new property values to the corresponding horizontal
    properties when the same characters are to be presented horizontally.

    If the main UCD file can hardly be changed, additional properties may be listed
    elsewhere in the file of extended properties.

    This archive was generated by hypermail 2.1.5 : Tue May 18 2004 - 18:08:19 CDT