Re: current version of unicode-font

From: Edward H. Trager (
Date: Fri Dec 03 2004 - 13:57:21 CST

  • Next message: Markus Scherer: "Re: Nicest UTF - string case mapping vs. UTF-8/32"

    On Thursday 2004.12.02 15:51:14 -0800, Richard Cook wrote:
    > On Thu, 2 Dec 2004, John Cowan xiele:
    > > Paul Hastings scripsit:
    > >
    > > > speaking of which, *are* there any open source fonts that come even
    > > > close to Arial Unicode MS?

    In the section on "Pan Unicode Fonts" on , I note:

        One disadvantage of pan-Unicode fonts are that glyphs from
        different scripts are often not aesthetically integrated across
        script blocks. Secondly, since it is nearly impossible for one or
        a few font designers to have been thoroughly schooled in the typographic
        traditions of numerous cultures and nations, glyphs in some of the
        script blocks may be of lower technical or aesthetic quality than glyphs
        in other script blocks. Finally, differences in national typographic
        traditions can lead to situations where glyphs which are adequate in one
        country appear odd or inappropriate in another country. For example,
        glyphs that are ideal for writing Arabic in Egypt are very likely not ideal
        for writing Urdu in Pakistan, even though both languages share the same
        basic alphabet ...

        (Visit the web site if you are interested in the rest of the commentary).

    Althought Arial Unicode MS is a very good Pan-Unicode font, it does most
    definitely suffer in that glyphs in certain script blocks are not aesthetically
    pleasing and may have less-than-ideal OpenType features for the placement of
    diacritics. For example, the glyphs in the Thai block in Arial Unicode MS
    are not nearly as aesthetically pleasting as the glyphs in fonts in the
    Thai National Font Project from NECTEC, so if you really want to read Thai,
    you are better off getting the Thai national font set.

    *That* thought leads us to another way to address
    the problem that Paul Hastings seems to be alluding to
    --the problem of having glyphs for every script one is going to come across--
    which is to install multiple fonts where each individual font is "optimized" for
    one or more script blocks that one is interested in. In this case, the answer
    to Mr. Hastings question can be answered:

       YES, there ARE Open Source (or otherwise liberally-licensed) fonts which are
       both aesthetically pleasing and are optimized for the correct placement of
       diacritics for a large number of different script blocks in Unicode.

    Of course the problem is, "where are they?" And the answer is, "they are
    all over the place!" To address this issue, I decided to winnow a lot of
    chaff from the grain and created the afore-mentioned resource:

    ... which I previously announced on one of the Linux internationalization
    mailing lists, but I think I neglected to announce it here (oops!).

    Take a look to see what is available. I put this resource together with the
    explicit aim of finding all of the open source and otherwise liberally-licensed
    fonts that I could for as many script blocks as I could. Of course it is not
    finished, but it is a start. (I have even thought of creating a little shell
    script which would let one download a wide selection of the fonts that are
    mentioned on the site. Of course, the problem is that people often change the
    version numbers and URLs of download sites, so a script has only a limited
    life time of utility ...).

    A final comment or two on the aesthetics of GNU Unifont and James Kass'
    CODE 2000:

       (1) GNU Unifont is not supposed to be pretty, it is a 100% utilitarian
       bitmap font. Period. There are areas where the bitmaps could be better, and
       I believe, if memory serves correctly, that James Su of Novell/SuSE did some
       work to improve some of the CJK section for SuSE's Linux
       distribution(s). Other sections of the font have some gaps, and the bitmap
       formats around which GNU Unifont was originally built do not support any of
       the OpenType, AAT or Graphite features needed for CTL layout of Indic and
       other scripts.

       (2) I have always believed that James Kass' objective with CODE 2000 was
       to *first* cover as much of the BMP as possible, and worry about aesthetic
       details later. For providing admirable coverage of Unicode for five
       dollars -- that sounds like a good deal to me, and I think CODE 2000 has
       served its purpose quite well. The global reach of the internet combined
       with the availability of Unicode-enabled operating systems from all the
       major vendors (Microsoft, Apple, Linux, Sun) is now spurring the development
       of numerous new fonts for native language computing in various scripts,
       -- and many of these are being released under open source licenses like the
       GPL or similarly liberal licenses such as what SIL has for Gentium and
       other fonts. The easy availability of these new fonts greatly reduces the
       need for Pan-Unicode fonts like CODE 2000 or MS Arial Unicode .

    -- 2004.12.03 Ed Trager

    > >
    > > In what, breadth of coverage or aesthetics? The GNU Unifont has very
    > > wide coverage though it is a bitmap font; James Kass's CODE 2000 and CODE
    > > 2001 probably have the widest coverage of any font, though it costs US$5
    > > to use them. Both of them IMHO are a tad on the ugly side.
    > In all fairness, the CODE 2000 font from James Kass is quite beautiful,
    > conceptually speaking. If the current execution is a tad ungainly here and
    > there, I ask 3 questions: (0) "What do you want for nothing (if you have
    > not yet paid the shareware fee)?"; (1) "What do you want for $5?"; and (2)
    > what do you want from a $5 shareware font that aspires to perfect coverage
    > of the *entire* BMP?
    > Code2000 is not open source, but Kass is remarkably responsive to user
    > input.
    > I urge everyone to download a copy of Code2000, and provide the developer
    > with feedback, both in terms of suggestions to improve the TrueType font,
    > and in terms of money to fund development.
    > James is doing some great work, using some relatively low-level
    > programming tools. In my experience (admittedly somewhat limited, since I
    > don't care about *everything* in the BMP) his font works where other
    > fonts, professional and amature, completely fail. If a font has the glyph
    > you need in any form, that's far better than having a glyph of last
    > resort, or no glyph at all.
    > Disclaimer: I have no commercial relation to Kass, and have received no
    > compensation for this endorsement. This review should also not be taken as
    > expressing approval of the shape of any glyph in the Code2000 font,
    > especially the Capital Letter J, which I think even Kass himself has
    > called "quirky at best". Note however that the Code2000 "hexagram" block
    > characters do look quite nice, and better yet, they work in Adobe
    > Illustrator CS, though no one (neither Kass nor Adobe) seems to know why
    > yet :-)

    This archive was generated by hypermail 2.1.5 : Fri Dec 03 2004 - 13:32:21 CST