Re: Generic base characters

From: Christopher Fynn (cfynn@gmx.net)
Date: Sun Jul 15 2007 - 08:48:13 CDT

  • Next message: Marcin 'Qrczak' Kowalczyk: "Ligatures, kerning, ZWNJ & OpenType"

    Peter Constable wrote:

    > From: unicode-bounce@unicode.org [mailto:unicode-bounce@unicode.org] On Behalf Of Christopher Fynn
    >
    >> The Microsoft OpenType shaping engine, Uniscribe, seems to
    >> automatically insert a dotted circle as a base character
    >> for isolated combining marks - and this behavior is outside
    >> of the control of the font developer.
    >>
    >> IMO it would be much better if font developers were
    >> responsible for defining their own sets of these base glyphs
    >> for combining marks - including the base for isolated
    >> combining marks - and their own lookups for rendering the
    >> resulting combinations.
    >
    > Uniscribe inserts a dotted circle glyph only when the author has not included a valid base character for the mark. Font developers are always responsible for their own lookups for rendering a mark glyph on a base. But there's nothing the font developer can do about the scenario in which an author fails to include a base.
    >
    > Perhaps you have in mind that a font developer should control what glyph is used in that situation, but I see a need, on the assumption that authors should, and normally are, explicitly intentional about what is in their document, and that Uniscribe's fallback rendering is just that: a fallback.

    I'd say first try and let the user (author) select the base they want to use;
    but, where the users choice is not supported by the font then fall back to any
    default which the font developer may have defined, and finally fall back to the
    dotted circle.

    With Uniscribe is there an issue here about using combining marks with base
    glyphs for characters outside the script to which the characters for the marks
    belong? I know for a fact that with some other OpenType rendering engines these
    need to be pre-defined in the shaping engine or the lookups won't work.

    >> As base glyphs one might want to include the dotted circle;
    >> non-breaking space or fixed width spaces such as em space or
    >> en space;
    >
    > Unicode specifies that combining marks in isolation -- with no visible base --
    should be combined with NO-BREAK SPACE, not any of the other space characters in
    the standard.

    OK, I was trying to allow some other spaces as well as NO-BREAK SPACE - I'll
    limit the lookups for no visible base to that character to make things
    conformant.

    >> Right now if I have a lookup in an OpenType font to place
    >> an isolated mark on the dotted circle Uniscribe also inserts
    >> dotted circle and I end up getting two doted circles...
    >> Not all OpenType shaping engines exhibit this behaviour.
    >
    > A bug, which can be looked at.

    If I want to get a dotted circle with all OT shaping engines I seen to have to
    include a specific lookup - but this results in two dotted circles in Microsoft
    applications. Whether this is due to Uniscribe or something else
    I'm not sure. For instance OpenOffice on Windows does not display this behavior
    - even though I think it takes advantage of Uniscribe when running on MS Windows.

    - Chris

    > Peter



    This archive was generated by hypermail 2.1.5 : Sun Jul 15 2007 - 08:53:38 CDT