GPG and charsets (Windows)
ajgpgml at tesla.inka.de
Sat Jan 4 15:43:02 CET 2003
Well, it seems that there is some general problem due to inconsistency with charsets under windows.
I'm using GPG1.2.1 on Win98se, and the console (command.com) here is using CP850 ("chcp" typed at the prompt returns "Active Codepage: 850").
The german translated texts are properly displayed with this charset, e.g. gpg --version displays "ü" as it should be and output redirected to disc and viewed from there with an Editor using standard cp1252 (aka. Latin1, ISO-8859-1) shows no "ü" but some strange char.
So far so good.
But the problems get started with passphrases and user-ids, as these stay with iso-8859-1 (as starting gpg with -vvv seem to prove), and the input from the console is not properly translated in that case.
This means: I have two possibilities: Don't use the console for --gen-key and let frontends do the translation (this would be ridiculous!) or live with Keys with "strange" UserIDs; again this cannot be really recommended, esp. as there is already Unicode support in the form of UTF-8 inside GPG!
So I would suggest to let either the gettext-output respect --charset-parameter too, or use the windows-apis to get the console-charset translated for use in GPG.
I would prefer the latter, as this wouldn't require the user to specify a charset in the gpg-conf.
Okay, this step might break some old passphrases, but you could offer a compatibility-option for those who are concerned.
More information about the Gnupg-devel