Re: [idn] The layers and character handling in DNS

[Eric Brunner-Williams in Portland Maine <brunner@nic-naa.net>]

Paul,

Is this a real requirement?

> we need to have authoritative and non-changing mapping tables for 
> every local charset

If real, is it strengthened or weakened by removing the word "mapping"?

I only ask because my experience at Sun and HP, and my peers at IBM and
elsewhere confirm, that historically vendors "cherry picked", going for
ROI (unit sales), prior to any code-set independent (or dependent if one
views the ASCII -> UTF8 substitution as just a code-set substitution)
i18n deployment. Vendor, Consortia, and National "authoritative" tables
were the norm, and were frequently modified.

I'm asking about cherry picking on tables.

Next, or to be confusing and go to your comment, with "mapping on",
previously, I only ask because the resolver supplied by the Chinese
Domain Name System (CDNS) discussed by Wenhui Zhang (let me know if
you haven't seen this mail, I thought it went to idn and provreg),
does convert GBK and Big5 encodings to UTF8. I assume something useful
(for some definition of utility) happens in some other CDNS supplied
service.

I'm asking about cherry picking on mappings.

If economically rational behavior is changed, then my prior experience
with cherry picking in the non-ASCII host system software market is a
limited guide.

Broadly, my question is "impossibility" relevant? Is the only alternative
to one path the one characterized as:

	"In an idealized universe where there was authoritative and
	 non-changing mapping tables for every local charset, ..."

I don't mind looking funny, idealized universii and formal reasoning are
run-of-the-mill for people who enjoy maths, but I'm puzzled by a course
so radically different from the historical provisioning of post-EBCDIC
writing system support, which has most of the CJ (I'm dead ignorant on K)
host market exchanging data like mad, with most traffic in a few encodings.

Incidently, until UCAS [1] went PDAM, about the time I started working on
i18n, my non-latin writing system didn't even have a prevailing encoding.

Thanks for your thoughts Paul,

kitakitamatsino (or U+15BD U+1455 U+15BD U+1455 U+14AA U+165F U+14C4)
Eric

[1] Dirk Vermeulen's work
http://www.egt.ie/standards/sl/n1441-en.html