[Ksummit-2008-discuss] proposal for discussion..

[Evgeniy Polyakov]

Hi Matthew.

On Mon, Aug 25, 2008 at 03:07:15PM -0600, Matthew Wilcox (matthew at wil.cx) wrote:
> I've certainly heard people say that netlink is hard to use from an
> application point of view.  Something about ugly hard-to-use macros.
> I've never tried it personally.  I've seen code that uses DBus and it
> doesn't look terribly pleasant either.  I suppose the real advantage is
> that D-Bus has wider adoption than netlink.

There are lots of helpers from pure userspace netlink ones
(like libnetlink) to quite high-level kernel abstractions
like connector and genetlink.

> > And don't get me wrong here. For some parts of the D-Bus protocol are  
> > really thing they are overdesigned and weird, but I have my 2 cents  
> > for a lot of protocols. Maybe someone just needs to define D-Bus as a  
> > RFC document. Would that make any difference? Would it make it a  
> > better specification?
> 
> Definitely.  The 'multiple independent implementations' thing is a
> big help.  By the time two or more people have fully implemented a
> spec, you've knocked most of the ambiguity out of it and have a group
> of people who can comment on what they actually meant when they wrote
> something down.

No one proposes to change DBUS itself or replace it with netlink. I just
want to say, that DBUS on top of netlink is a good way to move forward.
It may (or may not though) even allow to simplify some of its bits,
namely kernel side which works with group subsribtion, broadcasting
and definitely memory management, userspace code just needs to have a
wrapper to trim netlink headers.

> I'm currently doing a userspace implementation of the usbip protocol and
> there's no way anyone would have written that crazy protocol down, let
> alone get a second person to do an implementation.  Er, well, I suppose
> I'm an existance disproof of that.  But I intend to get the usbip code
> changed ;-)

Now suppose you will also want to implement ethernet and ip itself there
:)

-- 
	Evgeniy Polyakov