Re: intermediate summary (Re: filename encoding)

[Tomohiro KUBOTA]

Hi,

At Fri, 2 Feb 2001 15:32:16 +0000,
Edmund GRIMLEY EVANS <[EMAIL PROTECTED]> wrote:

> I think this table is misleading because it lists encodings instead of
> conversions. 

Then how about this:

-------------------+-----------------+----------------+---------------
                   |        1        |       2        |       3
-------------------+-----------------+----------------+---------------
encoding for       |own encodings    |own encodings   |locale
physical media     |                 |                |
   ^               |                 |                |
   | conversion    |locale-indep.    |locale-dep.     |(no conversion)
   | by kernel     |filesystem-dep.  |filesystem-dep. |
   v               |                 |                |
parameter for      |UTF-8            |locale          |locale
open()             |                 |                |
   ^               |                 |                |
   | conversion    |(no conversion)  |(no conversion) |(no conversion)
   | by libc       |                 |                |
   v               |                 |                |
parameter for      |UTF-8            |locale          |locale
fopen()            |                 |                |
   ^               |                 |                |
   | conversion    |locale-dep.      |(no conversion) |(no conversion)
   | by softwares  |filesystem-indep.|                |
   v               |                 |                |
encoding for       |locale           |locale          |locale
end user           |                 |                |
-------------------+-----------------+----------------+---------------
            [dep. = dependent, indep. = independent]

For example, according to 1., kernel has to have a converter
which is locale-indendent and filesystem-dependent.  Softwares
have to have a converter which is locale-dependent and filesystem-
independent.

In all ideas, end users can use their preferable encoding, because
'encoding for end user' is 'locale'.

---
Tomohiro KUBOTA <[EMAIL PROTECTED]>
http://surfchem0.riken.go.jp/~kubota/
"Introduction to I18N"
http://www.debian.org/doc/manuals/intro-i18n/
-
Linux-UTF8:   i18n of Linux on all levels
Archive:      http://mail.nl.linux.org/lists/