Lucas C. Villa Real
lucasvr at gobolinux.org
Tue Apr 28 12:03:05 PDT 2009
On Tue, Apr 28, 2009 at 3:49 PM, Havoc Pennington
<havoc.pennington at gmail.com> wrote:
> Only really makes sense to me if dbus-connection.c then used the
> timeout ... otherwise it will confuse everyone, I think. The timeout
> in the message could override the one passed to connection_send if
> connection_send timeout is -1, perhaps.
> But I don't know if it's worth it. I think you guys should just use
> dbus_message_set_data() really; it's quite efficient. Tough to imagine
> this is showing up in a real-world profile?
Problem with set_data() is get_data(), because we cannot know what the
slot id is in our send() wrapper function. We could use some internal
member of the message, or even the message itself, as a key to get the
slot id, but that's looking "way too much" for our simple problem. We
would like to avoid hashing if possible.
Indeed making connection_send() use that makes sense. We'll test that
change in the next few hours.
More information about the dbus