robtaylor at floopily.org
Thu Mar 9 07:56:49 PST 2006
Robert McQueen wrote:
> Thiago Macieira wrote:
>>Kalle Vahlman wrote:
>>>This definitely sounds like stuff that should be conveyed through the
>>>system bus, doesn't it?
> No, not really. How would you arbitrate this? The first user to log in
> takes the org.gnome.screensaver service name, and then the script is
> unable to request to lock anyone else's screen? If there was some way of
> enforcing (without enumerating usernames in the bus config, or something
> crap) that certain parts of the system bus namespace could be held only
> by programs running as a certain user, you'd be able to have some system
> component know it was talking to processes running as that particular
> user. Otherwise, the system bus isn't a replacement for a user-level
> bus, because you can't enforce which user you're talking to.
I'd say there should be a system daemon that the users sessions' screen
savers notify of their existence, with session bus address and service,
then acpi can just request the system service to signal all savers to lock.
Maybe this concept is generalisable.
>>The only feature I can think of for a user-level bus is to send broadcasts
>>or signals to all apps from that user.
> Er? No. How about services that represent or query resources which are
> conceptually belonging to the user, not to a particular session? Like
> the user's configuration settings, address book, connections to IM or
> VOIP servers...?
*nod*, but in that case Havoc's point that such a bus should be common
for all logins on all machines holds.
More information about the dbus