Reliability of the D-Bus broadcast messages

[David Zeuthen]

Hi,

On Tue, Mar 22, 2011 at 9:14 AM, Havoc Pennington <hp at pobox.com> wrote:
> There's also some weirdness if the memory limits in the system bus get
> involved, but the exact weirdness will probably change and this would
> only affect you on the system (not session) bus and only if you were
> sending very large or very numerous messages.

Does anyone have these bug references handy?

(Not having read said bugs I'd say that the semantics we want is
something that favors clients not servers ... e.g. we should let
signal listening peers simply assume that signals on the system bus
are never lost ... and simply rate-limit (maybe even disconnect) peers
that are too chatty (e.g. emitting large or numerous messages)... I
honestly thought this is the way it worked already, but I realize it's
probably more complicated than this given several security contexts
are involved....)

    David