Re: visible glyphs for U+2062 and similar characters

From: Jungshik Shin (
Date: Mon May 12 2003 - 04:22:58 EDT

  • Next message: Otto Stolz: "Re: EBCDIC code pages"

    On Sun, 11 May 2003, Michael (michka) Kaplan wrote:

    > From: "Jungshik Shin" <>
    > > Recently I found some fonts have visible glyphs for invisible characters
    > > (that I _guess_ are supposed to have no visual effect) such as U+2062
    > > (the glyph for U+2062 is dotted 'x' inside a dotted box). With these
    > > kind of fonts present, a bit naive(??) approach of searching for glyphs
    > > in all the fonts on the system would turn the 'invisible' to the visible.
    > > For instance, the following MathML snippet was rendered with
    > > a visible glyph (&InvisibleTimes; == U+2062) by Mozilla.
    > >
    > > <mi>a</mi><mo>&InvisibleTimes;</mo><mi>c</mi>
    > Yes, you mentioned this in your mails from a few days ago....

      You wrote as if I had sent two emails with more or less the same
    content a few days apart from each other. I didn't. Did you get two?

    > > My questions are:
    > >
    > > (1) Is a font to blame for having visible glyphs for U+2062 and
    > > similar characters. I think U+2062 and similar characters are different
    > > from ZWJ/ZWNJ and other 'control' characters that do have visual effect
    > > in such context as Indic scripts, Arabic script and expressing authorial
    > > intent about ligature in Latin and other scripts)
    > You did not mention specifically what font is doing this, which makes it
    > hard to claim that a font should be "blamed" for anything. What is/are the
    > font(s) and where did you get it/them?

      At that time, I hadn't checked it out yet. Now that I've checked it,
    it turned out that it's Code2000. It appears that James may have been a
    bit too aggressie to put visible glyphs for U+2062(⁢ : INVISIBLE TIMES)
    and a few other invisible math operators around U+2062. I think the best
    to do for them is to put invisible glyph (zero-contour, non-advancing
    or a very small width) for them by default and to have GSUB lookups to
    be turned on by 'font clients' in case it's desriable/required that
    they be rendered with visible glyphs. That way, 'naive' clients of
    the fonts would not render invisible characters with visible glyphs
    unknowingly. By 'naive', I mean passing a Unicode string containing
    U+2062-like characters without preprocessing (e.g. removing them )
    to a string-drawing API.


    This archive was generated by hypermail 2.1.5 : Mon May 12 2003 - 05:09:01 EDT