Re: Gedit plugin

[Pietro Battiston <me pietrobattiston it>]

Il giorno ven, 26/04/2013 alle 21.48 +0200, Stef Walter ha scritto:
> On 26.04.2013 16:38, Pietro Battiston wrote:
> > [...]
> >
> > When I use the DBus method EncryptText(), I am asked for a list of keys.
> > Assume I give some key of a friend of mine. When I try to decrypt the
> > resulting text, I would expect the process to fail - I don't have her
> > private key.
> > Instead, gpg decrypts it without problems, because the text is encrypted
> > both with my friend's key and _my_ key.
>
> I think if you have a default key it'll be automatically included in the
> recipients.

OK.

> > Fair enough, it can make sense, in principle, that I'm able to decrypt
> > any text I encrypted. But then, I would expect EncryptText() to also
> > accept an empty list of keys: it should then encrypt using only my key.
> > Instead, it refuses, pretending some recipients.
>
> Most likely a bug. This is implemented in libcryptui, which doesn't have
> much maintenance. but I would include a patch (and help you get started
> with building it) if you like.

Well, that's not crucial for me (and anyway I can't rely on it if it
doesn't get distributed, so that wouldn't help). I just wanted to be
sure I wasn't missing something.

Instead I have another question: of all the keys I get with the
"ListKeys" function of org.gnome.seahorse.Keys, some have a "raw-id"
looking like "04E29C73", some like "B0894B0104E29C73:2". All of my tests
seem to show that those of the first kind are able to encrypt, those of
the second are not
(org.freedesktop.DBus.GLib.UnmappedError.Org.gnome.seahorse.Error.Failed.Code1: Key is not a valid recipient 
for encryption). But can I reliably establish that I can drop all keys with the "raw-id" longer than 8 digits 
when asking the user for a key to encrypt? Or is there a better method to do this filtering?

Thanks again,

Pietro