charsets

[Werner Koch]

"Janusz A. Urbanowicz" <alex at bofh.torun.pl> writes:

> As far as I know GPG has some awareness of charsets other that US-ASCII. :-) 
> I'm by default using ISO-8859-2 (for Polish text) and here are three

I didn't know who needs another character set, so I have only koi8r
yet.  Okay, I'll implement latin-2 soon.

> 1) should I explicite specify any charset in GPG configuration ? Which one ?

The one you use on your machine. 

> 2) some of the ISO charsets are overlapping (like there are German umlauts 
> both in ISO 8859/1 and ISO 8859/2). Should there be any conversion tables ?

No.

> 3) by browsing the source I found that sometimes Unicode support kicks in.
> Shouldn't GPG operate on some Unicode encoding (like UTF-8) internally ? How

It does (partly).  The only place where we can't avoid to know about
the character set is when we are creating a new user ID.  the user ID
is expected to be in UTF-8 and this value gets hashed.  So we can't
change things later.  Displaying user IDs in the correct character set 
is not so important ;-)  

> is resolved such a situation when I'm signing/encrypting ISO 8859/2 text and
> the receiver expects it to be ISO 8859/1 ?

This is not within the scope of an encryption program.  GnuPG (and
OpenPGP) handles data transparently (if you don't min the strange
--textmode).


-- 
Werner Koch at guug.de           www.gnupg.org           keyid 621CC013