bus limits

[David Zeuthen]

On Sun, 2004-08-15 at 13:20 -0400, Havoc Pennington wrote:
> On Sun, 2004-08-15 at 12:48, David Zeuthen wrote:
> > Ok, the way I 'solved' this was to reconsider whether it's appropriate
> > at all to use D-BUS for the purpose of the point-to-point communication
> > that is sending messages from a hotplug helper spawned by the kernel to
> > a daemon. 
> > 
> > It pretty much isn't; it's private communication between two components
> > of the same software package (in this case hal), that's not entirely
> > what D-BUS is for
> 
> I do think this is a reasonable use of D-BUS, though I understand if
> you'd rather do something simple than try to answer the larger question
> about D-BUS limits behavior ;-)
> 

Heh :-) Well, maybe in point-to-point mode, but I was routing all my
messages through the bus, hence the limits issues. If messages can be
dropped it's a can of worms to do this way; here's what it looks like

      hal.hotplug              dbus-daemon-1              hald
          |           A             |                      |
          |------------------------>|           B          |
          |                         |--------------------->|
          |                         |                      |
          |                         |<---------------------|
          |<------------------------|           C          |
          |           D             |                      |

hal.hotplug invoked the method HotplugEvent service exposed by hald (I'm
not going to specify all the svc/if/obj's :-). Initially I just used
dbus_connection_send on a message with no_reply set and checked the
return value. Unfortunately this function doesn't return failure if a
limit have been reached. 

So I have to get a dummy reply from the service I'm invoking the call on
which first of all is a waste of bandwidth. However. We have four
messages in play here, A through D, anyone can get lost. I get a local
error if A fails (though that is not likely to happen), and 'The maximum
number of pending replies per connection has been reached' if B fails. 

What if C or D fails? Let's ignore D, that's not very likely (just like
A). C, on the other hand is just as likely to fail as B, the issue I
originally had. Well, hal.hotplug will get a timeout of some sort
waiting for a reply and then we have to try again. But this is bad since
hald would have to keep state to not process that method call again.

Cheers,
David