From: Philippe Verdy (verdy_p@wanadoo.fr)
Date: Sat Oct 22 2005 - 22:26:35 CST
From: "Richard Wordingham" <richard.wordingham@ntlworld.com>
> The Lao naming problems look like a type of typographical error, and
> definitely not problems due to using a Thai expert when a Lao expert was
> required.
A typographical error that swaps the names of two letters is definitely not
a typographical error if an expert was involved when validating the charts:
only two page to read, can it be that difficult to see for an expert?
May be experts were present during the encoding process, but not when the
final decision was made on the codepoint allocations, and not before the
document to publish in ISO/IEC 10646 was published. Then Unicode has
inherited from this decision.
Even if experts are involved I think that any new revision of Unicode of
ISO/IEC 10646 should go through a validation process that will ask experts
for final review of the documents, to check that no such error can occur,
and allow rapid correction of possible errors. The allocation map in Thai
and Lao is a complex grid with many holes that may finally contain such
errors, and it's obvious that such composition errors can appear, and
pathetic when this leads to an incoherent standard thart can't be changed
later due to stability.
Anyway, I don't think that the stability of character names is really
important. The stability of these normative names are only necessary for the
standardization process itself, but not for applications (including Unicode)
of the ISO/IEC 10646 standard, that simply use the final allocated
codepoints for all other references.
I have definitely stopped thinking that these normative names are useful in
any reference or application as identifiers (and their use in Unicode
regexps is just an unnecessary pollution, that does not make regexps more
useful and more expressive, given that equivalent codepoints can be used
instead). What remains is the role of these names as semantic hints, but
when these hints give false information, then it's best to use another list
of names, accurate for the language in which the names will be used (so yes
i support now the idea of creating separate lists of localized character
names, the existing names belonging to the technical "C" locale, and not
being even useful for the "root" locale).
This archive was generated by hypermail 2.1.5 : Sat Oct 22 2005 - 22:29:12 CST