On Fri, Nov 02, 2007 at 04:11:14PM +0100, Josselin Mouette wrote:
> Le vendredi 02 novembre 2007 à 15:46 +0100, Sjoerd Simons a écrit :
> > On Fri, Nov 02, 2007 at 03:25:06PM +0100, Josselin Mouette wrote:
> > > BTW, it would be nice if you could discuss this with the upstream
> > > developers, because currently they have no plans on doing otherwise.
> >
> > Basically the only environment variables you can expect when you've been
> > started by dbus is DBUS_* and the various early set environment variables
> > (LANG, PATH, etc).. When you've been started by a session daemon for an X
> > session, you can expect DISPLAY to be set correctly too..
> >
> > For everything else you will need some other way to recover them. In
> > seahorse's
> > case, it should implement some D-Bus api to recover the socket paths that
> > you'd
> > normally get from the GPG_AGENT_INFO and SSH_AUTH_SOCK environment
> > variables.
>
> The problem is not about seahorse. It is about processes spawned by the
> keybindings. They can be evolution and epiphany, but also, depending on
> the preferences, balsa, thunderbird, firefox, etc. Here the problem lies
> in a variable that is passed blindly by all these programs to gpg. And
> you can't "fix" all these applications, nor can you expect gnupg to
> implement a D-Bus API just to recover the path to a socket.
Then whatever is spawning these apps is what's the issue here.. Dbus activation
isn't a solution for everything
Sjoerd
--
Who does not trust enough will not be trusted.
-- Lao Tsu
