[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Inverted (=reflected) N

Robin wrote --

> > No they do not; they are just synonyms for glyphs.
> 
> as far as i'm concerned, the glyphs have nothing to do with the
> matter.

Which matter?  I agree that they are not what newmath.dtx is trying to
name (which is where this started) but I fear that they are the heart of
all matters as far as Berthold is concerned.

> the names represent some semantic notion;

agreed.

> in an ideal world
> each such semantic notion would be represented in the character set by
> a different code, but this doesn't happen any more.

I disagree: such smeantic notions have no need for numeric codes at all (apart
from low-level hash-codes).

> and unicode won't allocate separate names to the same code, whatever
> the semantic implication of the name.  i don't remember, but i'll bet
> there's a rule somewhere that says that character code standards
> aren't allowed to do that.

My reading of the Unicode standard is that it has little to do with
names; it has a conaonical 9and very long) name for each used slot but
it does not expect anyone to use that name and alows the use of any
name for a character, it is only concerened with what trandforamations
are made by software to the numerical codes: certainly it allows fonts
to use any name they like for glyphs since the standard is not concerened
with glyphs and I suspect it is not concerned with mathematical
entities either (but it will soon be asked to chnage its mind on that).

> 
> > > But there
> > > are also plenty of examples where one glyph stands for more than one
> > > character.
> > 
> > I am not sure that glyphs "stand for" characters at all.
> 
> quite so.

More agreement!

> a glyph is mapped, by the software that chooses to use the
> font that contains it, to a character code.

Only, possibly, by document reecognition software, I would think.  Why
else would one wish top convert from a medium-dependent visual
representation to a universal encoding for text-streams?

I think you mean it is converted to a numbered-slot; it is this
artefact of a particular highly specialised and non-portable
font-technology that drives so much effort to standardise font-slot
numbers these days.  Those who want to support only such global fixed
tables and unnecessary fixed slots may spend their time doing so but I
think it would be better to put one's time and money behind the more
modern and portable font technologies that support no-number glyph
specification (as well as numbered slots).

> that software responds to
> that character code by selecting the glyph.

Why does it not respond directly to the glyph name when selecting a glyph?

> it so happens that we often have conventional representations of
> characters,

as waht?  names, pixels, numbers?

> 
> it would surely lead to insanity if the names of glyphs were to be
> different in every font designed to be used with a given character
> set.  conventional representations and all that...

But the world is and always will be potentially insane in that kind of way.
If glyph designers do not want to think about the names then they are
free to use some standard (but I hope there will be more than one name
avaialble to them by such a process).

> 
> no, let's stick to the situation we have, where there are conventions
> (flouted of course by the big boys like microsoft) and we therefore
> have a finite task when we're building our software.

That's OK but please let those in the math community (for example) who
are interested decide what those conventions should be for math
entities, math glyphs and (when they arrive) math Unicode character
names; and please accept, as do a lot of groups who have spent a long
time thinking about this, that synonyms are natural and important.

> 
> but let's remember that, because of the shortage of slots in
> unicode[*], and that therefore one is forced in the real world to
> overload the semantics of characters by using aliases.


And an even bigger plea: do not ascribe more than the tiniest
of semantics (eg no more than a list of synonyms) to math entities or
glyphs, especially to those like \Ninv where the name question arises
(othewrwise we shall have to tell you not to use the uninverted N but
hundreds of different names depending on what it means; and we shall
have to invent conventions for all those and ...).

> 
> and of course, i have nothing to say about the naming of the glyphs
> required by maths.  i can see the value in making them consistent, but
> i can't get particularly excited about minor inconsistencies.

Well, some of us have lives which are that sad:-).

> [*] and the sorry fact of life that has SC2 taking a long time to
> produce standards -- 10646-1 _does_ `give them the freedom' to go away
> and do the right thing, in its intro about the architecture of 10646.

There are problems there but the numbers invloved in any conceivable
maths of the next 200 years will have no impact on the the size of
Unicode (we are talking about far less than 5000 slots).

The problems with Unicode as a global standard for text-streams are
more basic and philosophical, related to what text is in non-Western
cultures and whether it can be described by a fixed standard.


chris