Re: Why people still want to encode precomposed letters

From: John Hudson (
Date: Mon Nov 17 2008 - 01:06:05 CST

  • Next message: Julian Bradfield: "meaning of "two-level" in Han unification"

    Support for anchor attachment positioning of combining marks
    (mark-to-base and mark-to-mark) is the forward edge of current font
    development. It is something that has been widely implemented for some
    years for complex scripts, but only recently have the makers of fonts
    for European scripts begun to address this and, not surprisingly, there
    are some misconceptions and mistakes to be addressed. As the examples
    posted here over the last few days demonstrate, even those developers
    such as SIL and Microsoft who are providing mark attachment support for
    Latin script bases may not provide the same level of support for
    Cyrillic and Greek.

    I have recently been working on adding mark attachment positioning for
    all three of these scripts to a large font. It is an aspect of font
    development for which a good workflow has yet to be developed, and I
    hope to sit down with some colleagues next month, share observations on
    recent experience in this area, and see if we can come up with some
    recommendations for best practices, tool improvements, etc. Ensuring
    that all the scripts in a font get the same level of support will be on
    the agenda.

    I should add that the presence of precomposed diacritic characters in
    Unicode massively complicates the job of mark attachment positioning,
    requiring either glyph-level decomposition to base+mark sequences or
    definition of anchors on every precomposed diacritic glyph as well as
    simple bases. Having two mechanisms to achieve the same typeform display
    is always more work than having one.

    John Hudson

    Tiro Typeworks
    Gulf Islands, BC
    You can't build a healthy democracy with people
    who believe in little green men from Venus.
                        -- Arthur C. Clark

    This archive was generated by hypermail 2.1.5 : Mon Nov 17 2008 - 01:10:19 CST