RE: Representative glyphs for combining kannada signs

From: Keutgen, Walter (
Date: Fri Mar 31 2006 - 05:13:23 CST

Antoine, Kent,

I do not remember who is defending which point of view. I first want to recall my opinion as somebody not involved in the techniques of rendering. A dumb user.

1) German with strokes instead of dots for umlaut diaeresis.

Your question: Is one allowed to design such font?


The questions that you forgot before proceeding with such a project:

Is such a font nice? Matter of taste, ask the potential readers of the texts.
Is this font to be used with other languages?
If yes, are there confusions possible (with Hungarian double acute accent e.g.)
What about the tastes of these other language readers?

Is it allowed to write sometimes double dot diaeresis, sometimes double stroke?

No, not in the same font in the same text. You will confuse the readers.
If you want switch between documents, well look for a mechanism outside of the character encoding i.e. in the font.


2) "m2" is the plain text version of "m<SUP>2</SUP>", isn't it? The one a plain
text search will find in the corresponding HTML corpus, etc.

If your plain text encoding is basic ASCII AND "m<SUP>2</SUP>" is not part of a mathematical formula, yes.
If your plain text encoding is basic ASCII AND "m<SUP>2</SUP>" is part of a mathematical formula, the fallback will be
"m^2" or "m**2". But you might consider rearranging completely the formula for readability, which requires some mathematical knowledge.

If your plain text encoding supports U+00B2, then use just "m²".
If several different superscripts, carefully weigh the implications of applying different fallback methods versus a unique common one and perhaps you might conclude not to fall back at all.

If your plain text is in an Unicode encoding, there are further superscript decimal digits in [U+2070, U+2079]. Only few fonts support them it seems. Or should I better write, they are not easy to get onto the screen? Once the support is OK, you can keep at least numerical superscripts i.e. mathematical exponents, when switching to Unicode plain text.

Regarding the search, if the user cannot enter the superscript 2 into the search criterion or explicitly wants no differentiation, searching for "m2" should return "m<SUP>2</SUP>" and "m²".


3) Indian scripts

If I understand the reasoning, the question about German umlaut variant is to get an answer for an Indian script question.

This is a bad method if you want an "allowable or not" answer. What would make sense in the German printing tradition could not be accepted in the Indian and vice-versa.

The answers for German MAY give you HINTS on how to proceed for Indian.

Best regards


THIS COMMUNICATION MAY CONTAIN CONFIDENTIAL AND/OR OTHERWISE PROPRIETARY MATERIAL and is thus for use only by the intended recipient. If you received this in error, please contact the sender and delete the e-mail and its attachments from all computers.

-----Original Message-----
From: [] On Behalf Of Antoine Leca
Sent: Donnerstag, den 30. März 2006 11:55
Subject: Re: Representative glyphs for combining kannada signs

On Wednesday, March 29th, 2006 20:20Z, Kent Karlsson wrote:
> Antoine Leca wrote:
>>> For typeset modern German text DIEARESIS is consistently used
>>> (though most often via precomposed letters).
>> So, does it mean I am allowed to have/design a font that draws
>> diaeresis as two strokes (not dots),
> I'd say no.
>> for example to give some script-style look? Or am I not?
>> And if I am, am I furthermore allowed [...]
> No.

I am sorry, there is no chance I am able to make a point.

I am trying to explain a reasonment that goes point by point. You disagree
with the first point. Since the other are following the first, there is NO
NEED to discuss the further ones; at least until I amend the first in order
for us to agree.

I think I should really resist to post further if I cannot sucessfully
explain my premissae.

> There is no reason for that. I don't expect to ever see that
> seriously supported or seriously suggested.

For the record: I was thinking about OpenType mechanisms, particularly the
type 3 substitutions, as they are used in applications like Adobe InDesign
(perhaps the most widely deployed OT "application" after Uniscribe).

> "No" as in "no, I really meant plain text, no missing higher level
> markup".

"m2" is the plain text version of "m<SUP>2</SUP>", isn't it? The one a plain
text search will find in the corresponding HTML corpus, etc.

Or am I again missing something obvious here?

>>> Some characters do have overlapping glyph chapes.
>> And why could it *not* be the case for the Indian scripts?
> When the differences are considered (by at least some) significant
> enough not to be lost.

OK, at least this is clear: you consider the two possible renderings to be
different, so this translates to the need for everybody to be forced to
encode at plain text level some differences (i.e. spelling rules, which
remains to be fixed in some cases);
even if the original author does NOT consider there is a semantic
and even if those spelling difference may prevent or difficult practical

> I'm arguing a position that is not necessarily derived directly from
> "the book"...

Sorry, I did not catch this; this raging thread began with your
: This is already very clear, but apparently needs to be pointed out.
(on Thursday, March 23rd, 2006 22:42Z)


This archive was generated by hypermail 2.1.5 : Fri Mar 31 2006 - 05:17:16 CST