Re: Terminal Graphics Draft 2

From: Markus Kuhn (
Date: Sat Oct 10 1998 - 07:05:51 EDT

Frank da Cruz wrote on 1998-10-09 22:16 UTC:
> I appreciate that these characters are, in one sense, the same as the ones
> already in the Latin-1 section of Unicode.

Isn't it possible (and in fact most probable), that the poor designer of
this SNI font just had no clue about that these ordinal indicators are
usually superscripted, and therefore by accident included too large
glyphs of what really is the same character? The Unicode character/glyph
model clearly dictates here that the intention of what the characters
should have looked like matters and not what they actually do look like
by accident. Who knows, may be the vendor will fix the unfortunate
glyphs after this discussion in the next ROM revision, having learned
about the bug ...

The Unicode character/glyph model clearly requires that the designers of
mapping tables make themself well familiar with the intended application
of glyphs found in fonts.

> In the end, an SNI screen can show a feminine ordinal from its Latin-1
> character set and a "small letter a with underbar" from its Mathematisch
> character set side-by-side and their visual appearance will be strikingly
> different. And perhaps there is some application where this difference is
> significant.

I am fairly familiar with the mathematical notation, and there is
certainly no commonly used mathematical symbol pair "small letter a with
underbar" and "small letter o with underbar". It looks extremely likely
that the Latin-1 and the Mathematisch character set where just drawn by
different people, and they used different styles to draw the same
character. I can nicely imagine that the Mathematisch designer is asking
around colleagues during lunch for ideas on further characters, and some
Spanish collegue asks "Did you add our Spanish ordinal indicators? They
are used with numbers so should probably be in a math set." "No, what do
they look like?" "Just a small a and o with a line below." "Ok, will
do.", and having never seen these symbols herself, she just takes the
normal lower case (=small) a and o and puts a bar below. Sounds like a
pretty plausible genesis theory for these glyphs to me (especially
having worked for Siemens myself before, having seen undergraduate
students there drawing GUI icons for the control of machines which they
have never seen before, with surprising results before the first review
of the results.).

One alternative is always to get on the phone and try to reach the
people responsible for the original design of this terminal. Companies
like SNI usually keep tons of internal documents on subjects like this
around forever.

> Again, I'm not a fanatic about these particular characters, and as far as
> I can tell so far, nobody will notice or care if they are not encoded, but
> I want to be sure we have the selection process right, and this seems to
> be an excellent test case.

Agreed. I suggest that the criteria should say that accidentally
miss-shaped glyphs should not be forced as extra characters into


Markus G. Kuhn, Security Group, Computer Lab, Cambridge University, UK
email: mkuhn at,  home page: <>

This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:42 EDT