D-Bus optimizations

[Colin Walters]

On Wed, 2012-03-07 at 19:10 +0100, Rodrigo Moya wrote:
> On Mon, 2012-02-27 at 13:04 +0100, Rodrigo Moya wrote:
> > Hi
> > 
> > Just published a blog entry[1] about some recent work we have been doing
> > at Collabora to optimize D-Bus. So could you please have a look and
> > comment on it?
> > 
> > cheers
> > 
> > [1] http://blogs.gnome.org/rodrigo/2012/02/27/d-bus-optimizations/
> > 
> after some (failed) discussion on the linux-netdev mailing list, here's
> a 2nd blog post about the situation:
> 
> http://blogs.gnome.org/rodrigo/2012/03/07/d-bus-optimizations-ii/
> 
> so, could you please have a look at it and comment?

What I'm getting out of this is basically "here's a list of other
ideas".  We need to focus though on evaluating the tradeoffs. ZeroMQ for
example is totally different in practice than DBus.  Now I've never used
it myself, but I've heard good things about it.  My understanding is
that it's basically a framework for writing protocols, which is pretty
far below what DBus does.

DBus has some high level conventions like the type system, and also what
I think is most important, the RequestName system so that processes
can dynamically discover each other.

And over time people have been adding even higher level conventions
like the ObjectManager.

So when you say:

  "Use ZeroMQ, which is a library that, from a first look, provides the
stuff we need for D-Bus, that is multicast on local sockets." 

That's really not very useful.  Again the way to look at this is:

1) Are changes to applications required for a given proposal?
1a) Are there optional changes to applications that could make them
    faster?
2) Are changes to bindings required?
3) Are we changing both the system and session bus?
4) What is the estimated speedup?
5) How does the proposal affect longstanding dbus bugs like 
   https://bugs.freedesktop.org/show_bug.cgi?id=33606 ?