Re: wide-char is wide

[Trevor Daniels]

Robin Bannister Wednesday, March 25, 2009 4:17 PM


> Francisco Vila. wrote:
> > the right googleable word is Unicode, do you agree?
>
> Well, not fully.
> When I google for > unicode arabic percent
> I certainly end up at a relevant place
> http://www.fileformat.info/info/unicode/char/066a/index.htm
>
> But I am not done.
> I need to collect whatever it is \char needs,
> so I go looking for hexadecimals.
> There are lots of them in a nice table,
> and they are not all saying the same thing.
> This is where "UTF-32" could keep me straight.
>
>
> Back to NR 3.3.3
> > The following example shows UTF-8 coded characters being used
>
> My main point was: UTF-8 is wrong.

As this is describing the argument to \char you are
right.  \char takes the hexadecimal number representing
the Unicode code point. So the Arabic percent can be
x66a, x066a, x00066A, ...

> When you criticize UTF-32 as a replacement, are you
> implying that the next word "coded" is wrong too?

If we specify UTF-32 this would imply all the leading
zeros need to be expressed.  This is not required; any
valid hexadecimal representation of the integer is
acceptable.

> If so, I agree.
> The proper term is Unicode code point (mentioned at the top of 
> 3.3.3)
> and it is just an integer - no need to constrain how it is 
> represented.
> (But base 16 and the codespace slicing went hand in hand.)

No problem with that.

> So lets say
> > The following example shows Unicode code points being used

OK.  I agree "UTF-8" here is wrong.

> And further up, lets use this same term instead of
>  "Unicode escape sequence"  and  "Unicode hexadecimal code"

Happy to replace them but I prefer to use "Unicode
hexadecimal value".

Trevor