Bug#350275: [i18n] Hex codes instead of letters in d-i prompts

[Miroslav Kure <kurem@upcase.inf.upol.cz>]

Package: cdebconf


Hi,

when trying d-i daily installations in Czech, I can't enter accented
(iso-8859-2) characters into input boxes in d-i anymore. This happens
since the switch of d-i to unicode.

Instead of e.g. e with caron it displays "<EC>", which is hex code for
this letter in iso-8859-2. (236 in decimal or 11101100 in binary.)
It seems like the input layer is not somehow switched to utf-8 mode.


General facts about unicode on linux console (not specific to d-i):

1. Displaying of all latin2 characters works.

2. Entering latin2 characters works, if they are mapped to a single
   key (e.g. e with caron is on key "2", s with caron on key "3", ...)

3. Problem is with combined characters entered via dead key (e.g. you
   press ' (the dead key) and then a key which then displays a with
   acute). This known kernel console limitation in unicode mode.

In d-i I can't input even characters mentioned in point 2 (which works
on the console *after* the instalation).

The question of course is, if there is any d-i input, which requires
other characters than ACSII. And also if it will help our users that
they could write some accented characters, but not the others (which
require dead key, e.g. o with acute or d with caron in case of usual
Czech keyboard layout).

-- 
Miroslav Kure