DBus in the kernel?

[Colin Walters]

On Tue, Jan 5, 2010 at 11:41 AM, Ross Burton <ross at burtonini.com> wrote:
> On Tue, 2010-01-05 at 15:12 +0000, Michael Meeks wrote:
>> Last I recall, (from the a11y measurements that Rob did), that
>> like-for-like, d-bus was ~2x slower than ORBit2 for the same, simple,
>> point-to-point (ie. not via the bus) synchronous call.
>
> As a secondary data point, the last time I profiled eds-dbus sending
> thousands of contacts over the bus, message validation (specifically
> UTF-8 validation) was something like 20% of the runtime.
>
> Amusingly I attempted to switch the transport to arrays of bytes to skip
> the validation, but the array validation was actually even slower...

Did you have this commit?

commit d437d9202efd8190ec6405d04627b34cb47bcc86
Author: Jon Gosting <yukarionsen at gmail.com>
Date:   Mon Nov 10 23:29:05 2008 -0500

    Bug 18064 - more efficient validation for fixed-size type arrays

        * dbus/dbus-marshal-validate.c: If an array is fixed size,
        skip validation

    Signed-off-by: Colin Walters <walters at verbum.org>