IDNA Comparisons

[John C Klensin]

--On Thursday, July 16, 2009 16:50 -0400 Gervase Markham
<gerv at mozilla.org> wrote:

> Braille is, according to my limited understanding, a method
> for encoding  another character set in a form readable by
> people without sight. (The  version I know of encodes
> something pretty like ASCII, but perhaps in  other places in
> the world there are other versions encoding other  character
> sets.) I therefore can't see why one would want Braille domain 
> names; surely one would create the domain names in the
> character set in  question, and convert that to Braille on
> output (e.g. using Braille TTY)?

Yes.  And there is another part of the issue: as long as an
important, perhaps only of the most important, context in which
domain names are actually used is in URI or IRIs, we are _still_
stuck with the ASCII parts of those protocol elements.  Coding a
domain name in Braille while keeping the fixed parts in ASCII
poses far more difficult problems about context-switching than
"merely" having the protocol-specified parts of a URI or IRI in
basic Latin characters while the domain names and selected
elements of the tail are in some very different script.

If I want Braille output, then I want Braille output of the
entire URI or IRI (and probably the surrounding text), not just
of the domain name.  I also note, fwiw, that, while the active
users of Braille I've consulted would not want to be portrayed
as representing that community, they don't use Braille as an
input method from keyboards when dealing with computers -- ASCII
is a lot more efficient and compact, even if the output is
expected to be in Braille.  So, adding/permitting Braille in
IDNs would effectively just lead us into another mapping
requirement and I think we are still trying to avoid those when
they aren't necessary.

    john