Starting the kdbus discussions

[Lennart Poettering]

On Mon, 06.01.14 10:47, Simon McVittie (simon.mcvittie at collabora.co.uk) wrote:

> 
> On 03/01/14 23:18, Lennart Poettering wrote:
> > On Fri, 03.01.14 13:18, Simon McVittie (simon.mcvittie at collabora.co.uk) wrote:
> >>> As you can see: the first 16 bytes are 100% identical between dbus1 and
> >>> gvariant messages. Only beyond that the gvariant marshalling results in
> >>> different serialization.
> >>
> >> This seems perverse: if the header is compatible but the content isn't,
> >> then the message is not compatible, and as a result, there's no value in
> >> the header being compatible.
> > 
> > Well, the header already has a field for indinicating the "protocol
> > version", which we make use of here.
> 
> Ah - identical semantics, different contents. Sure, that's fine, but:
> 
> > dbus1 messages have protocol
> > version 1, gvariant messages have protocol version 2.
> 
> Please do not deploy "protocol version 2" anywhere until the D-Bus
> Specification describes it: this field (and anything else numeric,
 
Yeah. In order to make sure that distros don't accidentally turn on kdbus
in systemd and then maybe one day are incompatible with whatever we
decide on here we made this all strictly opt-in. In fact, you first have
to enable it in the configure switch, and then secondly on the kernel
cmdline. If you leave one out this stuffis not used, so that we have all
freedom to still change it before we agree on it.

> really) is not extensible except via D-Bus Specification updates. While
> kdbus is still being designed, a quick one-line patch that says
> "reserved for a GVariant-based serialization" (a lot like the current
> reservation of "m" for maybe types) would be enough.

Is this really necessary? I mean, this cookie is licked enough by
mentioning it here on the mailing list. And even if suddenly somebody
appears and super quickly introduces protocol version 2 and gets it
through the discussions and accepted, i have no problem at all with just
saying that gvariant gets protocol version 3 then...

Lennart

-- 
Lennart Poettering, Red Hat