Regarding gdm: I think why this is happening despite the fact that ConsoleKit upstream says that gdm shouldn't be starting CK until X has started is probably that we're carrying the patch in https://bugzilla.gnome.org/show_bug.cgi?id=593996. This queries CK for users before X starts.
However, it's wrong to assume that gdm is the only thing which starts ConsoleKit; other PAM sessions are quite liable to do so too, and it's entirely possible (if unlikely) for somebody to start a PAM session remotely while the X server happens to be starting. Thus, I argue that it makes sense to fix this in ConsoleKit independently of any changes in gdm.
Regarding gdm: I think why this is happening despite the fact that ConsoleKit upstream says that gdm shouldn't be starting CK until X has started is probably that we're carrying the patch in https:/ /bugzilla. gnome.org/ show_bug. cgi?id= 593996. This queries CK for users before X starts.
However, it's wrong to assume that gdm is the only thing which starts ConsoleKit; other PAM sessions are quite liable to do so too, and it's entirely possible (if unlikely) for somebody to start a PAM session remotely while the X server happens to be starting. Thus, I argue that it makes sense to fix this in ConsoleKit independently of any changes in gdm.