On Fri, Oct 13, 2000 at 04:40:56PM +0200, Karlsson Kent - keka wrote:
>
>
> > -----Original Message-----
> > From: Keld J�rn Simonsen [mailto:[EMAIL PROTECTED]]
> > Sent: Friday, October 13, 2000 2:41 PM
> > To: [EMAIL PROTECTED]
> > Subject: Re: Transliteration for use in UTF-8 locales
> >
> >
> > On Fri, Oct 13, 2000 at 10:08:25AM +0100, Edmund GRIMLEY EVANS wrote:
> > > Is it really a good idea to do transliteration in the locale? Surely
> > > it would be easier to have mbrtowc and wcrtomb be inverses and leave
> > > transliteration for iconv.
> >
> > Transliteration is cuturally dependent, eg cyrilic is very dependent
> > on which language you transliterate into.
>
> True, but I still don't think 14652 really has anything to say
> about transliteration, in particular not when converting from
> "wide" to "multibyte" representations. What is targeted is
> *fallback* rules, for use when the target encoding cannot
> represent a particular character. Such fallback rules may be
> inspired by transliteration rules, but that does not make the
> fallback rules into transliteration rules.
I agree that the 14652 specs were not intended for use in
C wide char to myltibyte conversion. We did not think of it in this way.
Maybe we should have and it is still doable in 15435.
> A transliteration (or transcription) API (having mappings
> from UCS string to (another) UCS string) is something else.
> I'm not sure if 14652 addresses that, and if it does, then
> it does so in a limited and somewhat strange fashion.
15435 actually have internal string to internal string conversion
(UCS-like) for transliteration. The current spec is limited to
non-context-dependent transliteration, which is the inherent meaning
of transliteration. Transliterating here mening transformation (trans)
of letters (literation). The transliteration is done with respect to
a repertoire that the resulting strin may contain.
Kind regards
Keld
-
Linux-UTF8: i18n of Linux on all levels
Archive: http://mail.nl.linux.org/lists/
