Re: LastResort BDF font updated

From: Adrian Havill (
Date: Mon Jun 23 1997 - 23:32:16 EDT

Unicode Discussion wrote:
> It seems to me that that it might be useful to have a second set of
> glyphs for LastResort purposes that didn't use a character from the
> script as the mnemonic so it would be easier to remember what they stood
> for and so it would be easier for the average unicode user to
> distinguish one from another.

Agreed. Mind you, I feel that the fonts target audience is not the
average developer that works with Unicode-- if I knew that the the
target audience was Unicode developers, I wouldn't bother with the
glyph, I'd just print "U+xxxx". After all, we've all memorized the
ranges for each set, yes. (^_^)

I can't expect a monolingual end user with no familiarity with Unicode
to differentiate between Tamil and Devangali, but I am kind of bothered
that many of the glyphs, especially those such as the CJK Compatibility
and the Kanbun, are confusing to Han users which are confortable with
the glyphs.

> Instead of two very different-looking glyphs such as
> "a" and "a" or katakana "ka" and hiragana "hi" or some other pair of
> glyphs less likely to trip up a non-Japanese-speaking developer.

How about Kanbun? Kanbun is a kind of "markup" that rides near the
original kanji, and the most common type of markup is a miniature
ideograph of a "One" or "Two" next to the original Chinese, indicating
the order the characters are to be read in for Japanese. The minature
raised "Chi" makes no sense at first glance. Only by looking it up I
finally went "Oh, THAT'S what it represents."

A much better glyph would be to have a frame outline, similar to the
technique used with the glyph for "Combining Diacritical Marks" etc.
with the kanbun in small to the side of it,, to emphasize that it's
kanbun, not kanji) This would be both clearer and more orthagonal, in my

> Since I'm sure the UTC put a lot of thought into the selection of just
> the right glyphs, and it's probably a done deal.

Did they [Technical Committee] select all the glyphs for the fonts? I
assumed they were based on the small sample in TUS 2.0. Some do not seem
to be well chosen. For example, the separate glyphs for Hiragana and
Katakana. There is no need for the separation, IMO. Just as there are a
few rendering devices left in the world that can't do lower case, their
are also a few rendering devices left that can do only one of the two
sets-- most of the time they are katakana only). But there are not
enough cases to justify the separation, just as their is no need to add
a lower case "a" to the unrenderables. A plain hiragana "ka" for "kana"
is enough. If the font is small enough to where holding both kana sets
is an issue, they probably won't be able to use the LastResort font
anyway, which is bigger than one kana set.

If a character set does hiragana and the display engine shows a
LastResort kana glyph, they can probably figure out that it's not
hiragana, and the "other" kana set, and vice versa.

On the flip side, the Japanese hankaku "ka" for half-width and
full-width forms is also confusing. What if the character is a full
width letter "A"? Is a Japanese character appropriate? Shouldn't the
latin fullwidth also use the latin "L" glyph?

My concern is that the target audience for the LastResort font, based on
current choice of characters, is the Unicode developer (which probably
would rather know the actual U+xxxx value anyway), which knows the
ranges and the glyphs and can do the leap of logic to figure out what
the glyph actually means.

I feel that a font is needed for applications that may be used by people
who are not in the know about Unicode, but who may be multilingual. As
it stands, some of the glyph choices for certain characters would throw
off even the people whose native language included the character.

In a quick ad-hoc and unscientific poll, I showed the font to the fifty
odd native Japanese in our company (via internal e-mail mailing lists)
without explanation and asked them to guess the meaning of glyphs with
Japanese characters in them.

None of fourteen who replied guessed all correctly as to what they
represented... most of the guesses being "superscript kanji" or
"furigana/rubi" for kanbun and "one character Japanese words" for CJK

Only one, a programmer who knows Unicode, got the CJK compatibility
glyph, but he cheated on the kanbun glyph-- he didn't know what it was,
so he figured it out by a process of elimination and by looking at the
adjacent characters to see where the set lay between in the Unicode map
and cross-referenced with TUS 2.0.

Adrian Havill <URL:>
Engineering Division, System Planning & Production Section

This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:35 EDT