On Wednesday 19 January 2005 1:26 pm, Pavel Krivanek wrote:
>         Czech localization works well on Linux too. Here's a patch. The
> only problem is with the clipboard inputs from the other programs but it's
> the issue of the Squeak VM. Clipboard operations from Squeak to Squeak are
> OK.

As I recall, there is support for declaring to the Unix Squeak VM that the 
input should be assumed to be in a particular character set. We use the 
values of various environment variables (and/or command line flags) to modify 
the arguments passed to the iconv routines.

So then you wouldn't have to have the (lossy) translation to MacRoman in 
between; you could just pass your own encoding up to the image. You'd 
probably have to come up with a different converter, though.

Have you tried changing the settings of (see "man squeak"):

SQUEAK_ENCODING

the name of the internal character encoding used by Squeak. Equivalent to 
giving the '-encoding' command-line option if set.


SQUEAK_PATHENC

the name of the character encoding used to construct file and directory names. 
Equivalent to giving the '-pathenc' command-line option if set.

SQUEAK_TEXTENC

the name of the character encoding used to copy/paste text from/to external 
applications. Equivalent to giving the '-textenc' command-line option if set.

LC_ALL
LC_CTYPE
SQUEAK_NOINTL
squeak also checks the environment for LC_ALL and LC_CTYPE. If either of these 
variables is set then support for international keyboards (including dead 
keys for diacritical marks) is enabled. To prevent this support being enabled 
even when one or both of these variables is set, use the '-nointl' option (or 
set SQUEAK_NOINTL in the environment). For example, to start squeak with 
support for dead keys on Spanish keyboards, with Latin-1 encoding of external 
characters and the default MacRoman internal font encoding, run squeak like 
this:

              export LC_CTYPE=es_ES
              export SQUEAK_TEXTENC=latin1
              squeak

-- 
Ned Konz
http://bike-nomad.com/squeak/