Announcing new version of the Qt4 bindings

Havoc Pennington hp at
Tue Feb 14 08:07:29 PST 2006

On Tue, 2006-02-14 at 11:42 +0100, Thiago Macieira wrote:
> The problem with fixing #1 would be that the API would change: the result 
> from the filters wouldn't be "handled / not handled", but would have to 
> indicate the reason why it didn't handle.

I missed this paragraph somehow. I think what Daniel says is probably
right. I would expect that if a handler is providing an object at the
exact object path as the message, then it should always return handled,
but may return an error (no such interface or no such method). But if
there's no object it should return NOT_YET_HANDLED, since someone else
may have registered the object.

Right now if all handlers return NOT_YET_HANDLED dbus-connection.c will
return UNKNOWN_METHOD, that should probably be changed to return

This seems to work, unless we want to support implementing an object
partially in one handler and partially in another, and that doesn't
sound important I would guess?


More information about the dbus mailing list