Re: (i)rregular number of columns in CJK code charts

From: Kenneth Whistler (
Date: Mon May 19 2008 - 14:23:59 CDT

  • Next message: Tjebbe van Tijen: "Re: (i)rregular number of columns in CJK code charts"

    Tjebbe van Tijen noted:

    > - some of the CJK code blocks that have more than 1 chart can have a
    > lesser number of columns for the last code chart
    > Like: 2F800 > 2FA1F CJK Compatibility Ideographs Supplement
    > which has 12 columns for the first 2/3 charts and 11 columns for the
    > last chard
    > This exception seems not to occur with the non CJK code charts.

    Actually, it does. Note, for example, the Ethiopic block, 1200..137F,
    which is split for display as 12-12 columns, rather than 16-8.

    > Just out of curiosity - because I am studying what could be called
    > the "spatial architecture of the Unicode" -
    > why is it that this in my view useful/nice regularity has its
    > exceptions?

    As Benjamin Scarborough surmised, it is simply the result of
    editorial decision to help balance out the display of columns
    for charts bigger than 16 columns wide. The unibook program
    that is used for chart formatting has various internal calculations
    to assist in this: some algorithmic, some hard-coded for particular
    situations. Note that the same program also decides when a
    chart is narrow enough that the names list for it can be
    formatted onto the right column of the same page, rather than
    on the succeeding page.

    These formatting decisions result from a combination of
    aesthetics for page display and an attempt to minimize the
    overall number of pages required for charts.

    I would caution not to draw architectural conclusions about the
    standard based upon such editorial contingencies that have more
    to do with book design than with character set architecture.


    This archive was generated by hypermail 2.1.5 : Mon May 19 2008 - 14:28:11 CDT