[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: support for multilingual Packages files?



On Sat, Jul 28, 2001 at 08:39:59AM +0900, Tomohiro KUBOTA wrote:
> 
> Various encodings are used in the world, such as ISO-8859-*, EUC-*,
> ISO-2022-*, KOI8-*, and so on so on.  English version of messages are
> used when translations are not available.  Thus, English version has
> to be avaiable in any locales.  Using non-ASCII character will limit
> dselect and so on to run only under UTF-8 locales, which is against
> (1) the current Debian situation that UTF-8 locales are not popular
> and (2) the future ideal Debian situation that should support not
> only UTF-8 locales but also all locales which the current Debian
> supports.  Even though I guess UTF-8 would be the only popular 
> encoding in the world in ten years, it does not mean that we can
> force users to use UTF-8 locales.
> 
> ASCII is the common subset of all encodings which Debian supports
> and which are very popular in the world.

I agree with this.
Though, this says nothing against using utf-8 in Packages file.

On Sat, Jul 28, 2001 at 03:12:56AM +0100, David Starner wrote:
> > I never said that you'd have to force users to use UTF-8. You could
> *store*
> > the text in UTF-8 (which for your ASCII requirement for English would work
> > too), and then dselect would either display the ASCII (no change) or will
> > translate the UTF-8 back to the charset used by the user's locale, which
> can
> > be ISO 2022 or whatever. It's definitely not difficult to do.
> 
> You miss the point. Part of the argument is that the English one should be
> in UTF-8, since many maintainer's names aren't in ASCII. Transliterating
> from full UTF-8 to any other character set correctly is impossible, due to
> lack of context, variable user expectations, and context sensitive

But most probably, it _is_ possible to recode those parts of document
user is interested in (since s/he speaks that language, that means
s/he has probably already set up proper console and locale for the
language [1]). The question, of course, is whether to include support
for iconv(3) into dselect or not. Or should we better concentrate on
improving utf-8 locales?

[1] not me, though, I would be interested in languages covered 
by a bit more than one locale. But I realize I am an exception :-)

> transliterations. Also, I want to be able to admin Debian from any terminal
> with ASCII support, and that includes possible running the Packages file
> through more (as dselect/ncurses may be broken, or worse, libc is broken.)

so? you run Packages through more. You see some garbage in some
maintainers' names, or in some descriptions. Do not tell me this is 
a problem (and besides, there ALREADY ARE 8-bit characters in Packages,
so this does not count as argument against utf-8 in Packages)

-- 
 -----------------------------------------------------------
| Radovan Garabik http://melkor.dnp.fmph.uniba.sk/~garabik/ |
| __..--^^^--..__    garabik @ melkor.dnp.fmph.uniba.sk     |
 -----------------------------------------------------------
Antivirus alert: file .signature infected by signature virus.
Hi! I'm a signature virus! Copy me into your signature file to help me spread!



Reply to: